Documente Academic
Documente Profesional
Documente Cultură
Laser Cooling
Fundamental Properties and Applications
edited by
editors
Preben Maegaard Galina Nemova
Anna Krenz
Wolfgang Palz
Wind Power
for the World
Published by
Pan Stanford Publishing Pte. Ltd.
Penthouse Level, Suntec Tower 3
8 Temasek Boulevard
Singapore 038988
Email: editorial@panstanford.com
Web: www.panstanford.com
For photocopying of material in this volume, please pay a copying fee through
the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923,
USA. In this case permission to photocopy is not required from the publisher.
ISBN 978-981-4745-04-8 (Hardcover)
ISBN 978-981-4745-05-5 (eBook)
Preface xiii
Index 453
Contents xiii
Preface
chapter reviews the theory and practice of this novel mode of laser
operation.
In Chapter 6 Stephen C. Rand compares three conceptually
different approaches to laser cooling of solids. He offers the unique
perspective that in solids where free translational motion is absent
there are nevertheless interactions analogous to the Doppler effect
that may be used to address the energy and momentum of collective
excitations of the medium. Moreover there are reasons to explore
them systematically. Two methods of cooling solids in particular,
which are reminiscent of interactions with free atoms, are analyzed
from this perspective to examine their detailed advantages and
disadvantages. While attempting to profit from the earlier history of
laser cooling in gases and to provide points of comparison between
different techniques, this chapter identifies an approach to cooling
that results in a temperature-independent cooling rate.
In the “traditional” model, which describes optical refrigeration
of rare earth–doped solids based on anti-Stokes fluorescence, a
cooling cycle is provided by resonance transitions between the upper
ground and lower excited levels of ions, phonon thermalization,
and subsequent radiative relaxation to the ground states. In such a
model, the rate of the cooling cycle is low because of the small cross
section of forbidden transitions between the levels of the same ion
configuration and the low population of the upper ground level. In
Chapter 7 Andrei Ivanov, Yuriy Rozhdestvensky, and Evgeniy Perlin
consider a cooling process with stimulated Raman adiabatic passage
for optical pumping of a rare earth–doped crystal taking into account
vibronic interaction at the Stark split sublevels of ion manifolds.
They show that coherent pumping by stimulated Raman adiabatic
passage leads to the deep laser cooling of rare earth–doped crystals
that have technologically attainable purity to date.
In Chapter 8 Azzurra Volpi, Alberto Di Lieto, and Mauro Tonelli
report on cooling efficiency measurements performed on various
Yb3+-doped fluoride single crystals grown in their laboratory. They
show that the investigation of the cooling process for increasing Yb3+
concentrations in LiYF4 (YLF) single crystals results in improved
cooling efficiency via significant decrease of detrimental parasitic
absorption. Energy transfer processes between Yb3+ ions and low
concentrations of rare earth ions impurities are investigated in this
chapter to analyze the effect of detrimental impurities, which partic-
Preface xvii
1.1 Introduction
Light as a tool for cooling matter was considered already in 1929
by P. Pringsheim [1]. The experimental realization of the laser [2]
paved the way for experiments on different laser cooling techniques.
Among these, Doppler cooling of dilute atomic gases is probably the
most common [3–5], first successfully investigated in the 1980s [6].
As an important step, this technique along with further evaporative
cooling allowed for the realization of Bose–Einstein condensation of
different atomic [7, 8] and molecular species [9] and has today grown
into a robust research area. An alternative approach for cooling is
anti-Stokes cooling in multilevel systems, culminating, for example,
in the cooling of solids. For the first time suggested in 1950 [10] and
the distance between the collisional partners grows and after its 1/e
natural lifetime—which is of the same order as in the unperturbed
case [23]—the electronically excited state decays under the
emission of a fluorescence photon. Since the energy levels at larger
interatomic distances approach those of the unperturbed radiating
atom A, the frequency of the emitted photon ωFl will be close to the
resonance frequency of atom A.
Figure 1.2 Cooling principle for the case of rubidium atoms subject to
high-density argon buffer gas, based on the calculated pseudo-
potentials [24]. If a rubidium atom collides with an argon atom,
absorption of far red-detuned radiation becomes feasible.
The formed quasi-molecule is excited, and the excited state
subsequently decays under emission of a fluorescence photon
mostly at larger interatomic distances, where the energy
levels correspond to those of the unperturbed alkali atom.
The fluorescence frequency ωFl is larger than the frequency of
the absorbed photon: an amount of energy in the order of the
thermal energy, kBT, is extracted from the sample. A cooling of
the gas mixture occurs.
n fl - n (1.1)
Pcool = a (n )Popt
n
The typical energy that can be extracted from the sample per
cooling cycle is of order of the thermal energy kBT. In the experiments
performed so far, we observe cooling powers Pcool in the order of
up to 100 mW [17, 23, 28, 29]. Note that it is important to choose
a detuning with a high-enough absorption probability a(ν). We are
aware that the collisional broadening plays a crucial role here. A
measurement of typical fluorescence data recorded with the dense
gas mixture will be shown in Section 1.4.1.
Experimentally, it turns out to be crucial that the used buffer
gas is of high purity to suppress the influence of concurring
heating effects, of which quenching is the most known. In the case
of quenching, the excitation energy is transferred to an atom of a
third species polluting the dense gas mixture when colliding with
the quasi-molecule. Relaxation can then occur via nonradiative
decay, and so energy is kept in the system, heating the gas. To reduce
possible quenching, an effect known to occur, for example, from
an admixture with nitrogen, we use a buffer gas with high purity,
namely argon gas preferably with a purity of 99.9999 (argon 6.0).
Residual heating can also be caused by energy pooling. In this
scenario, during the collision of two excited alkali atoms one of them
gets further excited. Its relaxation to the ground state implicates
the emission of a fluorescence photon with higher energy. For the
case of rubidium atoms, the reaction equation of that process can be
written as [30]:
Experimental Setup and Methods 7
For most experiments, both the incident laser beam and the
emitted fluorescence were spatially filtered with confocal pinholes.
The incoming laser beam was strongly focused to a beam waist of
approximately 10 μm. Fluorescence was then collected from that
area and analyzed by an optical spectrum analyzer. A sketch of that
setup is depicted in Fig. 1.4.
1
=
n0 Ú
— ^ n(r , t )dz (1.6)
1 dn
=
n0 Ú dr dz (1.7)
At this point we consider the deflection angle φ in radial
direction r to be small (dr/dz = tan φ ≈ φ). We derive an expression
for the temperature change as a function of the deflection angle by
reshaping Eq. 1.7:
1 dn
j=
n0 Ú dr dz (1.8)
1 dn dT
j=
n0 Ú dT dr dz (1.9)
Figure 1.5 Thermal deflection technique. The cooling beam and the
so-called probe beam are sent collinearly into the dense gas
mixture. The cooling zone induces changes to the gas density
and consequently to the refractive index. As a result, the
cooling zone acts like a graded index lens for the nonresonant
probe beam. By scanning the radial distance between the two
beams, the angular deflection φ can be measured as a function
of r.
1 Ê n2 - 1 ˆ
= = const. (1.10)
r ÁË n2 + 2 ˜¯
where ρ is the density of the medium. Assuming that the dense gas
mixture behaves like an ideal gas, its refractive index n is close to
unity. The temperature derivative of Eq. 1.10 can then lead to the
following expression for dn/dT:
dn (n - 1) (1.11)
ª-
dT T
Combining Eqs. 1.9 and 1.11, we obtain
(n - 1) dT
j=- Ú T dr
dz (1.12)
beams. In Fig. 1.6 the latter case is presented. The beam waist w0
of the probe beam is set to be smaller than that of the cooling laser
beam so that the entire cooling area can be resolved.
0.6
0.5
0.4
0.3
0.2
0.1
0.0
dT Ê Ê 2r 2 ˆ Ê 2r 2 ˆ ˆ
µ Á exp Á - 2 ˜ - exp Á - 2 ˜˜
(1.18)
dr Ë Ë w0 ¯ Ë 8Dt + w0 ¯ ¯
Laser Cooling Experiments on Dense Mixtures 19
Figure 1.10 Deflection angle (red spheres) of the He–Ne laser beam as a
function of its lateral offset to the cooling beam for a dense
rubidium-argon mixture (p = 230 bar, T = 620 K). A fit (solid
black line) on the measured data is performed following
Eq. 1.19. The temperature profile near the cell entrance derived
from the deflection data (solid blue line), as well as following a
heat transport model (dashed blue line) are in good agreement.
The maximum determined temperature drop is 66 K.
of Kennard [56] and Stepanov [57], such a universal relation has been
derived, applying for systems where efficient relaxation processes
lead to thermalization of both the ground and the electronically
excited state manifolds of the addressed transition. The so-called
Kennard–Stepanov relation can be written as follows:
Ê a (w ) ˆ h
ln Á ˜ = w + C(T ) (1.20)
Ë f (w ) ¯ kBT
where α(ω) and f(ω) are the frequency-dependent absorption
and emission line shapes, respectively, T is the so-called spectral
temperature and C(T) is a temperature-dependent constant. Its
derivation is based on the Boltzmann distribution of the occupation
probabilities within both the ground and the electronically excited
state manifolds and can be found elsewhere [58]. In the case of fully
thermalized state manifolds, the equation predicts a linear behavior of
the logarithm versus frequency, with a slope equal to ℏ/kBT. Linearity
deviations are mainly attributed to incomplete thermalization of the
excited state manifold, internal state conversion, and finite quantum
efficiency of the addressed transition. Extraction of the slope using
spectroscopic data of absorption and emission profiles can lead to
temperature determination of the sample. The method has already
been used for studies on dye molecules in liquid solutions [59,
60], solid state inhomogeneous systems [61, 62], and biochemical
photosystems [63], as well as thermalization studies of a photon
gas [64]. Even though an exact temperature extraction has not
been achieved in these experiments, mainly due to inhomogeneous
broadening [65, 66], the thermalization of the relevant states has
been verified through the linearity of the corresponding Kennard–
Stepanov plots. It is however expected that in high-purity gaseous
systems this contribution is strongly suppressed and an exact
temperature determination becomes possible. To this end, the
relation is applied to our dense rubidium–argon gas mixtures.
Before proceeding to the corresponding experiments, we note
the particular conditions under which the validity of the relation
is ensured. Even though the ground-state manifold of a transition
is quite often thermalized, for dilute gases under laser irradiation
this is not usually the case for the electronically excited state.
Most importantly, the thermalization rate is necessary to be much
faster than the radiative lifetime of the excited state, as only in this
Kennard–Stepanov Experiments 25
respectively, with the efficiency for laser heating (for blue detuning)
being larger than that observed for laser cooling (for red detuning).
This agrees with the intuitive expectations that to heat is easier than
to cool. The results can also be qualitatively understood from the
asymmetry between the absorption and the emission line shapes
for the pressure broadened gas predicted by the Kennard–Stepanov
relation.
1.6 Conclusions
We have reviewed experiments on redistributional laser cooling, a
technique currently applicable to dense alkali–rare gas mixtures.
The achieved cooling powers here are several orders of magnitude
larger than in Doppler cooling experiments for dilute atomic vapors,
and the typically achieved cooling efficiencies are in the order of
several percent.
The alkali transitions are strongly broadened by the high-
pressure gas. The linewidths are in the order of the thermal energy in
frequency units, that is, kBT/h. The cooling principle can be described
in the frame of a quasi-molecular treatment. Elastic collisions of the
optically active species with a surrounding perturber atom enable
optical excitation of the former by laser light, red-detuned to the
atomic resonances. After the collision, the interatomic distance
becomes larger, and the excited state decays under the emission of
a fluorescence photon after its natural lifetime. The photon energy
is near the difference between the ground and the excited state of
an undisturbed alkali atom and is thus higher than the energy of the
References 31
References
35. Jackson, W. B., Amer, N. M., Boccara, A. C., et al. (1981). Photothermal
deflection spectroscopy and detection, Appl. Opt., 20(8), pp. 1333–
1344.
36. Spear, J. D., Russo, R. E., Silva, R. J. (1990). Collinear photothermal
deflection spectroscopy with light-scattering samples, Appl. Opt.,
29(28), pp. 4225–4234.
37. Casperson, L. W. (1973). Gaussian light beams in inhomogeneous
media, Appl. Opt., 12(10), pp. 2434–2441.
38. Born, M., Wolf, E. (1999). Principles of optics: electromagnetic theory
of propagation, interference and diffraction of light/Max Born and
Emil Wolf with contributions by A. B. Bhatia, et al., 7th (expanded) ed.
Cambridge University Press, Cambridge.
39. Gordon, J. P., Leite, R. C. C., Moore, R. S., et al. (1965). Long-transient
effects in lasers with inserted liquid samples, J. Appl. Phys., 36(1), p. 3.
40. Zhu, Q., Pan, B., Chen, L., et al. (2010). Analysis of temperature
distributions in diode-pumped alkali vapor lasers, Opt. Commun.,
283(11), pp. 2406–2410.
41. Saß, A. (2014). Experimente zur stoßinduzierten Redistributions-
laserkühlung dichter atomarer Alkali-Edelgas-Mischungen und
Simulationen der zu erwartenden Temperaturverteilung, PhD thesis.
42. Shang-Yi, C. (1940). Broadening, asymmetry and shift of rubidium
resonance lines under different pressures of helium and argon up to
100 atmospheres, Phys. Rev., 58(12), pp. 1051–1058.
43. Daniel A. Steck, “Rubidium 85 D Line Data,” available online at http://
steck.us/alkalidata (revision 2.1.6, 20 September 2013).
44. Drummond, D. L. (1974). Potentials and continuum spectra of Rb-
noble gas molecules, J. Chem. Phys., 60(9), p. 3426.
45. Alcock, C. B., Itkin, V. P., Horrigan, M. K. (1984). Vapour pressure
equations for the metallic elements: 298–2500K, Can. Metall. Quart.,
23(3), pp. 309–313.
46. Mayer, H. (1931). Über eine neue Methode zur Messung kleinster
Dampfdrucke. Die Dampfdrucke von Quecksilber und Kalium, Z. Phys.,
67(3–4), pp. 240–263.
47. Vogl, U., Saß, A., Weitz, M. (2012). Laser cooling of dense rubidium-
noble gas mixtures via collisional redistribution of radiation. In
Epstein, R. I., Sheik-Bahae, M. (eds.) SPIE OPTO. SPIE, pp. 827508-1–
827508-8.
48. Saß, A., Forge, R., Christopoulos, S., et al. (2014). Laser cooling of
dense atomic gases by collisional redistribution of radiation and
References 35
62. Ihara, T., Maruyama, S., Yoshita, M., et al. (2009). Thermal-equilibrium
relation between the optical emission and absorption spectra of a
doped semiconductor quantum well, Phys. Rev. B, 80(3), p. 033307.
63. Zucchelli, G., Garlaschi, F. M., Croce, R., et al. (1995). A Stepanov
relation analysis of steady-state absorption and fluorescence spectra
in the isolated D1/D2/cytochrome b-559 complex, Biochim. Biophys.
Acta, 1229(1), pp. 59–63.
64. Klaers, J., Schmitt, J., Vewinger, F., et al. (2010). Bose-Einstein
condensation of photons in an optical microcavity, Nature,
468(7323), pp. 545–548.
65. van Metter, R., Knox, R. (1976). On the relation between absorption
and emission spectra of molecules in solution, Chem. Phys., 12(3), pp.
333–340.
66. Kawski, A., Bojarski, P., Kukliński, B. (2000). The local temperature
dependence of fluorescent centres in PVA films on the excitation
wavenumber, Z. Naturforsch. A, 55(6–7), pp. 653–655.
67. Ottinger, C., Scheps, R., York, G. W., et al. (1975). Broadening of the Rb
resonance lines by the noble gases, Phys. Rev. A, 11(6), pp. 1815–1828.
68. Christopoulos, S., Saß, A., Moroshkin, P., et al. (2015). A Kennard-
Stepanov relation study on redistributional laser cooling in dense
gaseous ensembles. In Epstein, R. I., Seletskiy, D. V., Sheik-Bahae, M.
(eds.) SPIE OPTO. SPIE, p. 93800M.
69. Kasha, M. (1950). Characterization of electronic transitions in complex
molecules, Discuss. Faraday Soc., 9, p. 14.
70. Lakowicz, J. R. (2006). Principles of Fluorescence Spectroscopy. Springer
US, Boston, MA.
71. Gelbwaser-Klimovsky, D., Szczygielski, K., Vogl, U., et al. (2015). Laser-
induced cooling of broadband heat reservoirs, Phys. Rev. A, 91(2), p.
023431.
72. Vdović, S., Sarkisyan, D., Pichler, G. (2006). Absorption spectrum
of rubidium and cesium dimers by compact computer operated
spectrometer, Opt. Commun., 268(1), pp. 58–63.
73. Simon, P., Moroshkin, P., Weller, L., et al. (2013). Towards redistribution
laser cooling of molecular gases: production of candidate molecules
SrH by laser ablation. In Epstein, R. I., Seletskiy, D. V., Sheik-Bahae, M.
(eds.) SPIE OPTO. SPIE, p. 86380C.
Chapter 2
Galina Nemova
Department of Engineering Physics, Polytechnique Montréal, C.P. 6079,
Succursale Centre-ville, Montréal (Québec) Canada H3C 3A7
galina.nemova@polymtl.ca
2.1 Introduction
Laser cooling is a physical process by which a system loses its
thermal energy as a result of interaction with laser light. The term
“laser cooling” is most often used to describe cooling and trapping
of atoms and ions to extremely low temperatures down to the nano-
Kelvin regime. The energy of molecules in gases, especially at low
temperatures, is more contained in translational degrees of freedom
then in vibrations. Laser cooling in such a system is based on the
Doppler’s effect. The basic idea is to red-shift the frequency of the
light to just below the resonance frequency of the atoms. In that case
the atoms moving toward the laser are blue-shifted into resonance.
They more strongly absorb photons then the atoms, which
move away from the laser and are red-shifted further away from
where E p is the pump laser radiation rate, Q is the heat flow power
delivered to the body, and E f is the fluorescence radiation rate. The
points over the symbols denote the derivative d/dt, where t is time.
The second law of thermodynamics requires that the entropy of an
isolated system does not decrease. It gives rise to the inequality
Q
Sf - - Sp ≥ 0 (2.2)
T
where Sp is the entropy of the pump source, Sf is the entropy of
fluorescence radiation, and T is ambient temperature. The entropy of
radiation (“photon gas”) is determined by the well-known formulae
of Bose statistic.
ln 2
S =kB ÚÚ c 3
[(1 + nn )ln(1 + nn ) - nn ln nn ]dn dW (2.3)
Figure 2.1 Entropy flux density of the radiation emitted from the surface
of a sample into a solid angle dΩ. The polar angle θ is measured
relative to the surface normal direction n.
The factor cos θ projects the unit area of the surface to be normal
to radiation. The energy flux density of the same radiation can be
described as
ln 2
E = ÚÚ c 2
hn nn cos q dn dW (2.5)
E (2.6)
T eff =
S
ln 2
E = 2 hn nn cos qDWDn (2.8)
c
respectively. Let us suppose that ether ∆ν or ∆Ω go to zero. Since
the energy flow density is nonzero the relation nν∆Ω∆ν in Eq. 2.8
must be finite, that is, nν tends to infinity (nν Æ ∞). Consequently,
Fundamentals of Laser Cooling in Rare Earth–Doped Solids 43
considering Eq. 2.7 for the entropy flow density one can conclude
that for monochromatic or unidirectional radiation fluxes the
entropy flow density vanishes. For the first time that was proved by
Landau in 1945. Using Eq. 2.6 one can conclude that the effective flux
temperature for monochromatic or unidirectional radiation fluxes
approaches infinity (Teff Æ ∞).
E f Tfeff Ê T ˆ
h= £ Á 1 - eff ˜ (2.9)
E p (Tfeff - T ) ÁË Tp ˜¯
where Tfeff and Tpeff are the effective flux temperature of the fluo-
rescence and the pump radiation, respectively. In the case of near-
monochromatic pump radiation the effective flux temperature
Tpeff Æ ∞ and, as one can see in Eq. 2.9, the maximum conversion ef-
ficiency depends only on the fluorescence effective flux temperature
Tfeff . This result for the first time was obtained by Weinstein in 1960
for electroluminescence [20].
and
Ê E11 - E 1 j ˆ
exp Á ˜
Ë kBT ¯
f1 j = (2.15)
3
Ê E11 - E1 j ˆ
Â
j =1
exp Á
Ë kBT ¯
˜
4 3
Â
i =1
f0i = Âf
j =1
1j = 1 (2.16)
nf =
Ú F (n )ndn (2.19)
Ú F (n )dn
where Φ(ν) is the photon flux density. The population of the lower
2F and the upper 2F5/2 manifolds can be calculated with the
7/2
relations
4 4
N0 = ÂN = Â f
i =1
0i
i =1
0i N0 , and (2.20)
Fundamentals of Laser Cooling in Rare Earth–Doped Solids 47
3 3
N1 = Â
j =1
N1 j = Âf
j =1
1 j N1 (2.21)
here λp = c/vp and lf = c nf are the pump wavelength and the
mean fluorescence wavelength, respectively. c is the speed of light
in vacuum. As one can see in Eq. 2.26 if the pump power intensity
reaches saturation IpSat = hn p (t f s e (n p )) the cooling efficiency, ηcool,
decreases in half of the cooling efficiency corresponding to the
weak pump intensity, which is described with Eq. 2.27. The cooling
efficiency described with Eq. 2.26 takes place in an ideal unrealistic
sample where reabsorption and the total internal reflection (TIR)
of radiated photons at the boundaries of the sample are not taken
into account. In actual samples used in experiments the process of
reabsorption, TIR as well as the cooperative effects, such as energy
Fundamentals of Laser Cooling in Rare Earth–Doped Solids 49
for example, Ωmax ≈ 203 cm–1 for a KPC crystal and Ωmax ≈ 1400
cm–1 for borate glasses. Since nonradiative decay is a source of heat
generation in the system only RE-doped hosts with Wnr << Wr are
suitable for laser cooling. The radiative relaxation rate of the first
excited state of RE ions is in the range Wr = 102–10
3 s–1 for a wide
3 –1
variety of hosts. Assuming Wr ≈ 10 s and using Eqs. 2.24 and 2.28
one can estimate that the emission quantum efficiency ηq > 0.99 can
only be reached in the samples with Ωmax ≈ < ΔE01/8 [31]. Here
ΔE01 is the energy gap between the manifolds involved into the
cooling cycle.
As one can see in Eq. 2.27, if the system is pumped at longer
wavelengths the cooling efficiency increases. However, as one moves
toward longer wavelengths the absorption cross section decreases
and the pump power absorbed by the sample decreases too. For this
reason, a figure of merit, Fcool(λ), for the cooling process has to be
considered.
Fcool(λ) = σabs(λ)ηcool(λ) (2.29)
The maximum Fcool(λ) can be used to evaluate the host material
for the laser cooling application.
Ú
of the sample. a r = a abs (n )F0 (n )dn is the average reabsorption
coefficient, which can be defined as the integrated product of the
absorption spectrum, αabs(ν), and the normalized fluorescence
intensity, F0(ν), here Ú F (n )dn = 1 . The fraction of emitted photons
0
As one can see in Eq. 2.30 the fluorescence lifetime, tf , increases
with the increase in the ion concentration and the size of the sample.
The fraction of photons, which leaves the sample and do not undergo
TIR is Pfree = 1 – cos(θc) = 1 – [1 – (1/n)2]1/2. For example, for the
Yb3+:ZBLAN sample with the refractive index n = 1.49 Pfree = 25.9%,
and for the Yb3+:YAG sample with the refractive index n = 1.833 Pfree
= 16.29%. Most of the anti-Stokes emitted photons undergo TIR. The
influence of TIR and reabsorption on solid-state optical cooling has
been investigated in details in Ref. [33] with a random walk model
developed in Refs. [34, 35]. The total escape efficiency or external
quantum efficiency in the system, where reabsorption and TIR are
taken into account, is
(hq )N {exp( -a r L)/ n - [1 - (1 n)2 ]1 2 + 1}
hext = (2.31)
1 - {[1 - (1 / n)2 ]1 2 - exp( -a r L) n}(hq )N
where N = G(L/lave)2 is the number of emission effects, lave is the
step size or mean free path, that is, the distance corresponding to
a probability for reabsorption of one half, that is, exp(–lave·αr) = 0.5.
52 Laser Cooling in Rare Earth–Doped Glasses and Crystals
hSamp
min
ª
(
es B Tr4 - Ts4 ) (2.36)
hcools a (n p )NT IpSat
Figure 2.4 illustrates the dependence of the threshold GEF of the
sample, hSamp
min
, simulated with Eq. 2.36 on the ion concentration if
the final equilibrium temperature of the sample is Ts = 80 K or 100 K.
power. As one can see in Eq. 2.33 the temperature of the sample, Ts,
decreases as the cooling power, Pcool, increases. The cooling power
Pcool = –ηcool∙Pabs is proportional to the absorbed pump power. The
pump power absorbed in the sample on the length Labs is
Pabs = Pp (1 - e -a abs Labs ) (2.37)
(a) (b)
Figure 2.6 Nonresonant cavities formed by dielectric mirrors deposited
on the sample (a) and by external mirrors (b).
line shapes are generally the processes that shorten the lifetimes
of the energy levels involved in the transition, as an example, the
Heisenberg uncertainty relationship relating time and energy.
This contribution can be considered as the natural linewidth for a
transition. The homogeneous spectral line shape can be described
with the area-normalized Lorentzian profile
2Dn L
gL (n ,T ) = (2.38)
p ÈÎ( Dn L ) + 4(n - n0 )2 ˘˚
2
where ν0 is the mean transition frequency (Fig. 2.7a). ΔνL is the full-
width at half-maximum (FWHM) given by the relation [50, 51]
1 1
Dn L = + (2.39)
2pt r pT2
where T2 is the dephasing time. As the temperature decreases the
homogeneous transition linewidth ΔνL is observed to narrow [53].
At very low temperatures (up to 40 K), the homogeneous linewidth
is observed to follow a linear increase law with temperature (i.e.,
ΔνL ≈ T) at higher temperatures (up to 300 K), the increase is nearly
quadratic (i.e., ΔνL ≈ T2) [51, 52]. The one-phonon direct broadening
is associated with liner temperature dependence; the two-phonon
Raman scattering is associated with a T2 low at higher temperatures
[54]. Each dopant ion occupies a unique site in the host material and
experiences a unique Stark splitting. The structural disorder in a
host is a source of inhomogeneous broadening, which is generally
well described by a Gaussian distribution (Fig. 2.7b). In a perfect
(unrealistic) crystal all dopant ions are subjected to an identical
crystal field and as a result are free from inhomogeneous broadening.
(a) (b)
Figure 2.7 Homogeneous broadening (a) and inhomogeneous broadening
(b).
Laser Cooling in Rare Earth–Doped Glasses and Crystals 59
Figure 2.10 The progress toward net laser cooling in RE-doped glasses and
crystals with a record temperature drop for each host material.
Temperature Measurements 69
2.6.1.1 Thermocouples
In 1968, Kushida and Geusic used a thermocouple to observe heat
mitigation in a Nd3+:YAG sample. It was attached to the sample. Later
a direct measurement of the temperature of the entire sample well
insulated in an evacuated chamber has been done with a fine wire
thermocouple in a number of experiments [45, 63, 76]. Conduction
of heat through the thermocouple wires and absorption of the
fluorescence are obstacles on a way high-quality measurement,
which can be avoided with noncontact techniques.
slow and in addition is not immune to inherent noise from CCD dark
current and electronic readout. The resolution of ~20–100 mK can
be reached with the DLT method.
2.7 Conclusions
This chapter is focused on relatively comprehensive consideration
of the laser cooling process in RE-doped glasses and crystals. Laser
cooling of solids is interesting not only from the fundamental point of
view as a process of light and material interaction but its application
in portable space-based sensor systems as well. An understanding
of laser cooling of solids is very important for development of an
all-solid-state cryocooler, which is a compact, reliable device free
from vibrations. It is free from moving parts and fluids. It can be
used for cryogenic refrigeration of semiconductor devices in a
number of applications that have tight constraints for weight and
size, for example, airborne and space-based detector systems.
The operation of semiconductor detectors is based on electron
excitation. The incident photon excites electron from the valence
to conduction band inducing an electric current, which has to be
measured. But random thermal excitations (thermal noise) can also
excite electrons from the valence to conduction band resulting in the
“dark” current that limits the performance of detectors. Cryogenic
refrigeration of detectors can sufficiently reduce the thermal noise
and increase the detector sensitivity. Liquid helium, methane, and
ethylene ammonia can be used for cooling down to 150–50 K. But
such coolers have a large weight, a finite supply of refrigerant, and a
short lifetime (~1 year). Widely used thermoelectric coolers (TECs)
with longer lifetimes can reach only ~175 K. Cryogenic refrigeration
free from vibrations is important for high-performance IR cameras
74 Laser Cooling in Rare Earth–Doped Glasses and Crystals
References
18. Nemova, G., Kashyap, R. (2010). Laser cooling of solids, Rep. Prog.
Phys., 73, pp. 086501-1–20.
19. Zhang, J., Li, D., Chen, R., Xiong, Q. (2013). Laser cooling of a
semiconductor by 40 kelvin, Nature, 493, pp. 504–508.
20. Weinstein, M. A. (1960). Thermodynamic limitation on the conversion
of heat into light, J. Opt. Soc. Am., 50, pp. 597–602.
21. Chukova, Yu. P. (1969). Thermodynamic limit of the luminescence
efficiency, JETP Lett., 10, pp. 294–296.
22. Chukova, Yu. P. (1971). Thermodynamic limit to the efficiency of
broadband photoluminescence, Bull. Acad. Sci. USSR Phys. Ser., 35, p.
1349.
23. Chukova, Yu. P. (1974). Influence of excitation-line characteristics
on efficiency of spectral conversion of energy by ions of trivalent
neodymium in yttrium-aluminum garnet, Bull. Acad. Sci. USSR Phys.
Ser., 38, p. 57.
24. Chukova, Yu. P. (1976). The region of thermodynamic admissibility of
light efficiencies larger than unity, Sov. Phys. JETP, 41, pp. 613–616.
25. Sheik-Bahae, M., Epstein, R. I. (2009). Laser cooling of solids, Laser
Photonics Rev., 3, pp. 67–84.
26. Hehlen, M. P., Sheik-Bahae, M., Epstein, R. I. (2014). Solid-state optical
refrigeration, in Handbook on the Physics and Chemistry of Rare Earths,
Vol. 45, pp. 179–260.
27. Luo, X., Eisaman, M. D., Gosnell, T. R., (1998). Laser cooling of a solid by
21 K starting from room temperature, Opt. Lett., 23, pp. 639–641.
28. Gosnell, T. R. (1999). Laser cooling of a solid by 65 K starting from
room temperature, Opt. Lett., 24, pp. 1041–1043.
29. Kiel, A. (1964). Multi-phonon spontaneous emission in paramagnetic
crystals, in Third International Conference on Quantum Electronics
(Grivet, P., Bloembergen, N., eds.), pp. 765–772 (Columbia University
Press, New York).
30. Riseberg, L. A., Moos, H. W. (1968). Multiphonon orbit-lattice relaxation
of excited states of rare earth ions in crystals, Phys. Rev., 174, pp. 429–
438.
31. Hehlen, M. P. (2009). Design and fabrication of rare-earth doped laser
cooling materials, in Optical Refrigeration: Science and Applications of
Laser Cooling of Solids (Epstein, R. I., Sheik-Bahae, M., eds.), pp. 37–74
(Wiley, Weinheim, Germany).
32. Auzel, F. (2004). Upconversion and anti-Stokes processes with f and d
ions in solids, Chem. Rev., 104, pp. 139–173.
References 77
46. Hoyt, C. W., Hasselbeck, M. P., Sheik-Bahae, M., Epstein, R. I., Greenfield,
S., Thiede, J., Distel, J., Valencia, J. (2003). Advances in laser cooling of
thulium-doped glass, J. Opt. Soc. Am. B, 20, pp. 1066–1075.
47. Seletskiy, D. V., Hasselbeck, M. P., Sheik-Bahae, M. (2010). Resonant
cavity-enhanced absorption for optical refrigeration, Appl. Phys. Lett.,
96, pp. 181106-1–3.
48. Seletskiy, D. V., Hasselbeck, M. P., Sheik-Bahae, M., Epstein, R. I., Bigotta,
S., Tonelli, M. (2008). Cooling of Yb:YLF using cavity enhanced resonant
absorption, Proc. SPIE, 6907, pp. B9070–B9079.
49. Seletskiy, D. V., Melgaard, S. D., Bigotta, S., Di Lieto, A., Tonelli, M.,
Epstein, R. I., Sheik-Bahae, M. (2009). Demonstration of an optical
cryocooler, CLEO/IQEC 2009, post deadline submission.
50. Weber, M. J. (1985). Laser spectroscopy of glasses, Ceram. Bull., 64, p.
1439.
51. MacFarlane, R. M., Shelby, R. M. (1987). Homogeneous line broadening
of optical transitions of ions and molecules in glasses, J. Lumin., 36, pp.
179–207.
52. Yen, W. M., Brundage, R. T. (1987). Fluorescence line narrowing in
inorganic glasses: linewidth measurements, J. Lumin., 36, pp. 209–220.
53. Siegman, A. E. (1986). Lasers (University Science Books. Mill Valley,
CA).
54. Pellegrino, J. M., Yen. W. M., Weber, M. J. (1980). Composition
dependence of Nd3+ homogeneous linewidths in glasses, J. Appl. Phys.,
51, pp. 6332–6336.
55. Lei, G., Anderson, J. E., Buchwald, M. I., Edwards, B. C., Epstein, R. I.
(1998). Determination of spectral linewidths by Voigt profiles in Yb3+-
doped fluorozirconate glasses, Phys. Rev. B, 57, pp. 7673–7678.
56. Hoyt, C. W., Sheik-Bahae, M., Epstein, R. I., Edwards, B. C., Anderson, J.
E. (2000). Observation of anti-Stokes fluorescence cooling in Thulium-
doped glass, Phys. Rev. Lett., 85, pp. 3600–3603.
57. Bowman, S. R., Mungan, C. E. (2000). New materials for optical cooling,
Appl. Phys. B, 71, pp. 807–811.
58. Epstein, R. I., Brown, J. J., Edwards, B. C., Gibbs, A. (2001). Measurements
of optical refrigeration in ytterbium-doped crystals, J. Appl. Phys., 90,
pp. 4815–4819.
59. Poulain, M., Poulain, M., Lucas, J. (1975). Verres fluores au tetrafluorure
de zirconium proprietes optiques d’un verre dope au Nd3+, Mater. Res.
Bull., 10, pp. 243–246.
References 79
73. Patterson, W., Bigotta, S., Sheik-Bahae, M., Parisi, D., Tonelli, M., Epstein,
R. I. (2008). Anti-Stokes luminescence cooling of Tm3+doped BaY2F8,
Opt. Express, 16, pp. 1704–1710.
74. Mungan, C. E., Buchwald, M. L., Edwards, B. C., Epstein, R. I., Gosnell,
T. R. (1997a). Laser cooling of a solid by 16 K starting from room-
temperature, Phys. Rev. Lett., 78, pp. 1030–1033.
75. Mungan, C. E., Buchwald, M. I., Edwards, B. C., Epstein, R. I., Gosnell,
T. R., (1997b). Internal laser cooling of Yb3+-doped glass measured
between 100 and 300 K, Appl. Phys. Lett., 71, pp. 1458–1460.
76. Rayner, A., Hirsch, M., Heckenberg, N. R., Rubinsztein-Dunlop, H.
(2001). Distributed laser refrigeration, Appl. Opt., 40, pp. 5423–5429.
77. Heeg, B., Stone, M. D., Khizhnyak, A., Rumbles, G., Mills, G., DeBarber, P.
A. (2004). Experimental demonstration of intracavity solid-state laser
cooling of Yb3+:ZrF4-BaF2-LaF3-AlF3-NaF glass, Phys. Rev. A, 70, pp.
021401-1–4.
78. Seletskiy, D. V., Melgaard, S. D., Epstein, R. I., Di Lieto, A., Tonelli, M.,
Sheik-Bahae, M. (2011). Local laser cooling of Yb:YLF to 110 K, Opt.
Express, 19, pp. 18229–18236.
79. Seletskiy, D. V., Melgaard, S. D., Epstein, R. I., Di Lieto, A., Tonelli,
M., Sheik-Bahae, M. (2012). Precise determination of minimum
achievable temperature for solid-state optical refrigeration, J. Lumin.,
arXiv:1203.5967
80. Nemova, G., Kashyap, R. (2014). Laser cooling in Yb3+:YAG, J. Opt. Soc.
Am. B, 3, pp. 340–348.
81. Bigotta, S., Parisi, D., Bonelli, L., Toncelli, A., Tonelli, M., Di Lieto, A.
(2006). Spectroscopic and laser cooling results on Yb3+-doped BaY2F8
single crystal, J. Appl. Phys., 100, pp. 013109-1–7.
82. Bigotta, S., Parisi, D., Bonelli, L., Toncelli, A., Di Lieto, A., Tonelli, M.
(2006). Laser cooling of Yb3+-doped BaY2F8 single crystal, Opt. Mater.,
28, pp. 1321–1324.
83. Bigotta, S., Di Lieto, A., Parisi, D., Bonelli, L., Toncelli, A., Tonelli,
M. (2007). Single fluoride crystals as materials for laser cooling
application, Proc. SPIE, 6461, pp. 64610E–646110E.
84. Nemova, G., Kashyap, R. (2012). Laser cooling with Tm3+-doped oxy-
fluoride glass ceramic, J. Opt. Soc. Am. B, 29, pp. 3034–3038.
85. Soares de Lima Filho, E., Krishnaiah, K. V., Ledemi, Y., Yu Ye-Jin,
Messaddeq, Y., Nemova, G., Kashyap, R. (2015) Ytterbium-doped glass-
ceramics for optical refrigeration, Opt. Express, 23, pp. 4630–4640.
References 81
86. Boccara, C., Fournier, D., Jackson, W., Amer, N. M. (1980). Sensitive
photothermal deflection technique for measuring absorption in
optically thin media, Opt. Lett., 5, pp. 377–379.
87. Jackson, W. B., Amer, N. M., Boccara, A. C., Fournier, D. (1981).
Photothermal deflection spectroscopy and detection, Appl. Opt., 20,
pp. 1333–1344.
88. Garcia-Adeva, A. J., Balda, R., Al Saleh M., Garcia-Revilla, S., Fernandez,
J. (2011). Local internal and bulk optical cooling in Nd-doped crystals
and nanocrystalline powders revisited, Proc. SPIE, 7951, pp. 795106-
1–10.
89. Garcia-Adeva, A. J., Balda, R., Al Saleh, M., Garcia-Revilla, S., Sola, D.,
Fernandez, J. (2012). Optical cooling of Nd-doped solids, Proc. SPIE,
8275, pp. 827502-1–14.
90. Ghasemkhani, M., Seletskiy, D. V., Sheik-Bahae, M. (2011). Sensitive
thermal reflectance measurement for laser cooling applications, Proc.
SPIE, 7951, pp. 79510J-1–6.
91. Ghasemkhani, M., Seletskiy, D. V., Sheik-Bahae, M. (2012). Polarization-
resolved optical metrology for noncontact thermometry, Proc. SPIE,
8275, pp. 82750J-1–5.
92. Farley, C. III, Reddy, B. R. (2011). Mach-Zehnder interferometric
measurement of laser heating/cooling in Yb3+:YAG, Proc. SPIE, 7951,
pp. 79510I-1–8.
93. Rayner. A., Friese, M. E. J., Truscott, A. G., Heckenberg, N. R., Rubinsztein-
Dunlop, H. (2001). Laser cooling of a solid from ambient temperature,
J. Mod. Opt., 48, pp. 103–114.
94. Condon, N. J., Bowman, S. R., O`Connor, P. O., Quimby, R. S., Mungan, C.
E. (2009). Optical cooling in Er3+:KPb2Cl5, Opt. Express, 17, pp. 5466–
5472.
95. Seletskiy, D. V., Melgaard, S. D., Bigotta, S., Di Lieto, A., Tonelli, M., Sheik-
Bahae, M. (2010). Laser cooling of solids to cryogenic temperatures,
Nat. Photonics, 4, pp. 161–164.
Chapter 3
3.1 Introduction
Laser cooling has been successfully demonstrated on rare earth–
doped glass [1–3] and CdS semiconductor crystals [4]. (For recent
general reviews of progress in solid laser cooling, see, for example,
Refs. [5, 6].) These advances prompted considerations on how
laser cooling of solids may be implemented on a practical scale
for applications. In particular, it was noted (2010 Office of Naval
Research STTR Solicitation and private communications with Dr.
Deborah Van Vechten, ONR) that the possibility of transferring
waste heat as photons to remote locations is a desirable advantage
of optical refrigeration. These considerations have motivated us to
Tm-doped
cooling fiber
Heating
Tm-doped
fiber laser PL Source PF
Heat dump
Figure 3.1 Schematic of the fiber cooling geometry. A fiber laser pumps
the fiber cooling segment, which is attached to a heat source.
The up-converted photoluminescence light can, in principle, be
piped arbitrary distances before being dumped and converted
to heat.
dN 2 g I g
= s 12 N1 − 1 N 2 inc − RN 2 + w2 N 3 − 3 N 2 e − d Eu /kBT
dt g2 hn g2
(3.2)
86 Progress toward Laser Cooling of Thulium-Doped Fibers
dN3 Ê g ˆ
= -RN3 - w2 Á N3 - 3 N2e -d Eu /kBT ˜ (3.3)
dt Ë g2 ¯
Ntotal = N0 + N1 + N2 + N3 (3.4)
|3> g3 = 4
δEu w2
|2> g2 = 5
hν Wrad Wnr
|1> g1 = 6
δEg w1
|0> g0 = 7
4.3×10-4
10
a(10-3m-1)
2.2×10-4
5 a(dB/cm)
Pnet (W/cm3)
0.0
0
0 1 2
Optimal pump irradiance (109 W/m2)
Figure 3.4 Schematic drawing of the cross section of the fiber. The core
is a Tm-doped tellurite fiber of radius 52.5 mm, and the outer
radius of the cladding is 62.5 mm.
0.30 1.0
Absorption (black) [10-20cm2]
0.25 0.9
Emission (red) [arb. units]
0.8
Normalized intensity
0.20 0.7
0.6 285 K
0.15
0.5 290 K
0.4 295 K
0.10
0.15 300 K
0.3 305 K
0.05 0.2 310 K
0.1 315 K
0.00 0.0
1400 1500 1600 1700 1800 1900 2000 1600 1640 1680 17201760 1800 1840 1880 1920
Wavelength [nm] Wavelength (nm)
Figure 3.5 Left panel: the measured photoabsorption cross section of the
Tm-doped glass (black) and the calculated emission spectrum
(red) at 300 K of the pumped fiber according to Eq. 3.14. The
absorption cross section has been smoothed before being used
in Eq. 3.14. Right panel: the calculated emission spectra for
various temperatures. All the curves are normalized to 1 as
their peak values. Reprinted with permission from Ref. [12],
Copyright 2013, Society of Photo Optical Instrumentation
Engineers.
Temperature drop (T–T0) [K]
0
0
Cooling power [W/cm3]
–10 –2
–20 –4
–30
–6
–40
–8
0 –2 –4 –6 –8 0.0 0.1 0.2 0.3 0.4
Figure 3.6 Left panel: calculated temperature drop of the fiber as a function
of its cooling power Pnet (T0 = 300 K). Right panel: cooling
power of the pumped fiber as extracted from the experimental
emission spectra (black squares) and as calculated by the
four-level model in Section 3.2.1 (red curve). Reprinted with
permission from Ref. [12], Copyright 2013, Society of Photo
Optical Instrumentation Engineers.
Theoretical Developments 93
Thermistor, resistors
Metal heat 0.5 mm
spreader
V-groove 0.5 mm
Fiber
3 cm
Silica glass base
y
x Doped core
z Cladding
Fiber cross section
Pump turned on
Core temperature (K)
Cooling by fiber
450 307.41
400 307.405
350 307.4
300 307.395
Y (microns)
250 307.39
200 307.385
150 307.38
100 307.375
50 307.37
0 307.365
0 50 100 150 200 250 300 350 400 450 500
X (microns)
300
307.25
250
307.2
200
250
150 307.15
100 307.1
50 307.05
0 307
0.5 1 1.5 2 2.5
Z (cm)
can be seen from Fig. 3.11, the total loss due to metal impurities
absorption in the fiber glass is about 0.002 dB/cm around 1.9 mm
spectral region where optical cooling processes occur in Tm-doped
glass.
0.020
0.018
Impurities Absorption (dB/cm)
0.016
0.014
0.012
0.008
loss in glass
0.006
0.004
1ppm Fe+2 abs
0.002
1ppm Co+2 abs
0.000
0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 2.1
Wavelength (micron)
Figure 3.11 Loss due to metal impurities in Tm+3-doped tellurite glass:
1 ppm Cu, 1 ppm Fe, 1 ppm Co, and total loss due to actual
impurities in glass. Reprinted with permission from Ref. [12],
Copyright 2013, Society of Photo Optical Instrumentation
Engineers.
100 Progress toward Laser Cooling of Thulium-Doped Fibers
As is well known, the hydroxyl group –OH also has big impact
on optical cooling due to the following reasons: (i) Energy transfer
from Tm ions in excited levels to –OH proceeds via a nonradiative
decay, which generates heat in the system, and (ii) contributions to
the fiber loss in general (absorption and scattering, for example)
can also generate heat. In this work, we have achieved reducing –OH
absorption in glasses TE-II from 75–90 dB/m normal tellurite glass
to ~8 dB/m at 3.3 mm, as shown in Fig. 3.12.
100
90 TWD-S4
TD-S23
80
TD-S24
OH Absorption (dB/m)
70 TD-S25
TD-S26
60 TD-S27
50
40
30
20
10
0
2250 2500 2750 3000 3250 3500 3750 4000 4250 4500
Wavelength (nm)
Figure 3.12 –OH absorption in Tm-doped tellurite glasses TD-S23 to TD-
S27 compared to standard glass; TDS-25 glass (blue) has been
used for fabrication of cooling fiber Tm-CF2. Reprinted with
permission from Ref. [12], Copyright 2013, Society of Photo
Optical Instrumentation Engineers.
on the overall cooling efficiency. At the same time, the cooling fiber
configuration should be able to guide the fluorescence light in the
inner cladding. Second, the fiber configuration must satisfy the
thermal conditions that can maximize the transfer of heat from the
heating sample to the fiber core where cooling effect takes place.
We have fabricated three Tm-doped tellurite glass cooling fibers.
All fibers have core glasses doped with 2 wt% Tm concentration.
The main differences between those fibers are the glasses with the
different levels of TM impurities and –OH absorption. To improve
the potential cooling performance in the fibers, the different fiber
sizes are also designed and fabricated to improve pump absorption
and also coupling losses with passive fibers. The specifications of the
first cooling fiber (Tm-CF1) are: 2% Tm-doped tellurite glass (TE-
I); core/clad size (diameter): 50/125 mm. The glass TE-I has been
fabricated with raw materials having a standard level of impurities.
However, the –OH absorption is reduced to ~8 dB/m compared to
75–90 dB/m of standard glasses. The second cooling fiber (Tm-CF2)
is 2% Tm-doped tellurite glass (TE-II); core/clad size: 105/125 mm.
The glass TE-II has been fabricated with highly purified materials
having low impurities (all TMs are less than 1 ppm). Figure 3.13
shows the two cooling fibers Tm-CF1 and Tm-CF-2.
Figure 3.13 Images of two Tm-doped tellurite glass cooling fibers Tm-CF1
(left) and Tm-CF2 (right). Reprinted with permission from Ref.
[12], Copyright 2013, Society of Photo Optical Instrumentation
Engineers.
0.036
0.032
Absorption dB/cm
0.028
0.024
0.020
Tm-CF1
0.016
0.012
0.008
Tm-CF2
0.004
0.000
1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2
Wavelength (micron)
Figure 3.14 Impurities loss in cooling fibers Tm-CF1 and Tm-CF2.
Heaters(resistors)
Coupling fiber
Figure 3.15 The experiment setup and some important parts in the cooling
experiment. The cooling fiber is underneath of the metal
(silver) plate.
The experiments are repeated with each pump power after about
5 hours. In the experiment, the initial temperature in the vacuum
chamber and the sample are set at temperature 26.5°C. When the
pump turns on, the temperature measured by the thermistor at
the metal plate rises quickly to 29.5°C (P = 1.6 W, @ 1940 nm).
Keeping the pump power unchanged, the temperature versus
time was measured and recorded from 17:30 to 22:30, as shown
in Fig. 3.16. Closer inspection of the results in Fig. 3.16 indicates a
temperature difference between the sample and the inside of the
vacuum chamber that is consistent with cooling. The temperature in
the sample drops more than 0.6°C. We note that the input power is
1.6 W and the pump residue is about 0.25 W, meaning that the fiber
absorbs (including loss) 1.35 W of pump power continuously during
that time. We have carried out cooling experiments using the second
cooling fiber Tm-CF2 and the same configuration as in Fig. 3.15. The
temperature evolution for the fiber Tm-CF2 is similar to Fig. 3.16,
but the magnitude of temperature in the sample drops more than
1.2°C versus 0.6°C in first fiber Tm-CF1.
105/125
Beam combiner NA=0.2 Beam combiner
TEC
1.45 W -5C
Normalized Intensity
1.0 T = -10 C
0.9 P = 2.55 W -10 C
Normalized Intensity
P = 2.90 W
0.8
0.7
0.6 From room temperature
0.5
Temperature drops
0.4
to - 10C (P=2.55 W)
0.3
0.2
0.1 P=2.90 W
0.0 T < -10 C
1600 1640 1680 1720 1760 1800 1840 1880 1920
Wavelength (nm)
experiment theory
1.0 1.0
0.90 W
0.9 1.45 W
Normalized Intensity
2.05 W
Normalized Intensity
1.0 +24 C
1.0 0.18 W 0.9 +12 C
0.9 0.45 W +5C
Normalized Intensity
Normalized Intensity
0.65 W 0.8 0C
0.8 0.90 W -5C
1.20 W 0.7 -10 C
0.7
0.6
0.6
0.5
0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0.0 0.0
1600 1640 1680 1720 1760 1800 1840 1880 1920 1600 1640 1680 1720 1760 1800 1840 1880 1920
Wavelength (nm) Wavelength (nm)
Figure 3.21 shows the results for the 1940 nm pump laser. Again,
the results are consistent with the interpretation that net optical
108 Progress toward Laser Cooling of Thulium-Doped Fibers
Acknowledgments
Elements of this work were sponsored by the Office of Naval Research
(ONR) under contract/grant STTR N00014-11-C-0477, ONR Public
Release Distribution Statement A, “Approved for public release;
distribution is unlimited.” Additional support was provided by the
Air Force Office of Scientific Research (AFOSR) under contract/grant
FA9550-11-0048 and by TRIF Photonics. We thank G. Hardesty for
his contributions to our theoretical effort.
References
7. Nguyen, D. T., Shanor, C., Zong, J., Tian, W., Yao, Z., Wu, J., Weiss, J.,
Binder, R., Chavez-Pirson, A. (2011). Conceptual study of a fiber-optical
approach to solid-state laser cooling, Proc. SPIE, 7951, p. 795109.
8. Epstein, R. I., Sheik-Bahae, M., eds. (2009). Optical Refrigeration
(WILEY-VCH Verlag, Germany).
9. Hoyt, C. W., Hasselbeck, M., Sheik-Bahae, M., Epstein, R. I., Greenfield,
S., Thiede, J., Distel, J., Valencia, J. (2003). Advances in laser cooling of
thulium-doped glass, J. Opt. Soc. Am. B, 20, pp. 1066–1074.
10. Mungan, C. E., Buchwald, M. I., Edwards, B. C., Epstein, R. I., Gosnell,
T. R. (1997). Laser cooling of a solid by 16 K starting from room
temperature, Phys. Rev. Lett., 78, pp. 1030–1033.
11. Gosnell, T. R. (1998). Laser cooling of a solid by 16 K starting from
room temperature, Opt. Lett., 24, pp. 1041–1043.
12. Nguyen, D. T., Thapa, R., Rhonehouse, D., Zong, J., Miller, A., Kwong,
N. H., Binder, R., Chavez-Pirson, A. (2013). Towards all-fiber optical
coolers using Tm-doped glass fibers, Proc. SPIE, 8638, p. 86380G.
13. McCumber, D. E. (1964). Einstein relations connecting broadband
emission and absorption spectra, Phys. Rev., 136, pp. A954–A957.
14. van Roosbroeck, W., Shockley, W. (1954). Photon-radiative
recombination of electrons and holes in germanium, Phys. Rev., 94, pp.
1558–1560.
15. Zimmermann, R. (1987). Many-Particle Theory of Highly Excited
Semiconductors (Teubner, Leipzig).
16. Kwong, N. H., Rupper, G., Binder, R. (2009). Self-consistent T-matrix
theory of light-absorption and luminescence, Phys. Rev. B, 79, p.
155205.
17. Richtmyer, R. D., Morton, K. W. (1967). Difference Methods for Initial-
Value Problems, 2nd Ed. (Interscience Publishers, New York).
18. Nguyen, D. T., Zong, J., Rhonenhouse, D., Yao, Z., Hardesty, G., Kwong,
N. H., Binder, R., Chavez-Pirson, A. (2012). All fiber approach to solid-
state laser cooling, Proc. SPIE, 8275, p. 827506.
19. Hehlen, M. P., Epstein, R. I., Inoue, H. (2006). Laser cooling in rare-earth
doped solids: the next step, Proc. SPIE, 6130, p. 61300N.
Chapter 4
4.1 Introduction
Since the first demonstration provided by Epstein et al. [1] two
decades ago, laser cooling of solids or optical refrigeration has
been developed systematically [2–4], and recent advances have
claimed cryogenic cooling in highly purified Yb3+-doped fluoride
crystals [5–8]. The simple energy level structure together with the
wide energy gap imparts Yb3+ ion–doped sample an extremely high
quantum efficiency, making Yb3+ ion an ideal choice for cryogenic
optical refrigeration. Nevertheless, other trivalent rare earth ions,
such as Tm3+, Er3+, and Ho3+, attract increasing interest due to two
main reasons. For one thing, the operation wavelength region for
optical refrigeration can be expanded by adopting these ions, which
could be technically important in practical applications. For another,
the ASFC process. The rest of the 5I7 Ho3+ ions de-excited and transfer
their energy to the neighboring 5I7 Ho3+ ions, causing population
accumulation at the third excited state (5I5) with host phonon
absorption (solid green arrows). Comparison between the energy
mismatch and the host phonon energy (for YLiF4 crystal host, ħω ≈
450 cm–1) indicates that about two host phonons are annihilated for
each ETU process (see Table 4.1).
(solid purple and brown arrows), eliminating heat from the host
(upconversion-assisted cooling), or nonradiatively (dashed gray
arrows on the right), introducing extra heat to the host.
dt
dn2
= -( b20W20 + b21W21 )n2 + b32W32 n3 (4.1b)
dt
dn1 I
= s abs (n )(n0 - n1 ) + b31W31 n3 + b21W21 n2
dt hn (4.1c)
-W10 n1 - 2p11 (C )n
2
1
 n = CN (4.1d)
i
i =0
where εij is the energy gap between i and j levels. By using Eqs.
4.4–4.7, the approximate expression of the cooling efficiency can be
written as
n 10
¢ = hext
hcool ¢ habs - 1 (4.10)
n
where
-1
habs = [1 + a b s abs (n0 s - n1 s )] (4.11)
is the pump intensity– and frequency-dependent absorption
efficiency [5]. In Eq. 4.10, we have introduced the equivalent external
quantum efficiency
-1
¢ = (1 + W10¢ ,nr W10 ,r )
hext (4.12)
118 Laser Cooling of Solids around 2.07 Microns
where
¢ = W10,nr + p11 (C )n1 s (4.13)
W10,nr
-1
Ê i -1
ˆ
xij = b ij Wij ,nr Á
Ë
 b ik Wik ˜ (4.15b)
¯
k =0
Conventional Laser Cooling of Holmium-Doped Fluoride Crystals 119
Figure 4.2 (a) Absorption and emission cross section (π-pol.) of Ho3+
5I manifold; emission cross section (π-pol.) of Ho3+ 5I (b)
7 6
and 5I5 (c) manifolds; and (d) absorption and emission cross
section (π-pol.) of Tm3+ 5F4 manifold in YLiF4 host matrix
at room temperature [20, 23] (the Tm3+ spectrum will be
needed in Section 4.4). The dashed lines indicate the average
fluorescence wavelengths.
Conventional Laser Cooling of Holmium-Doped Fluoride Crystals 121
Figure 4.3 (a) The cooling power density and (b) the cooling efficiency of
a Ho3+:YLiF4 crystal as functions of pump intensity and dopant
concentration. The pump wavelength is 2070 nm, and the
background absorption is 4.0 × 10–4 cm–1.
MCPD
MCPD
ODC ODC
Figure 4.5 The maximum cooling power density (MCPD) and the corre-
sponding optimal dopant concentration (ODC) verses pump
wavelength at background absorptions of (a) 4.0 × 10–4 cm–1
and (b) 4.0 × 10–5 cm–1.
dt hn e
dn2
= - A2 + B2 (4.22c)
dt
dn1 Ig
= s g (n g )(n0 - n1 ) - A1 + B1 - 2p11 (C )n1 (4.22d)
2
dt hn g
 n = CN (4.22e)
i
i =0
Ai = Â b W n (4.24)ij ij i
j =0
is the sum of all decay terms from the ith level (composed of radiative
and nonradiative transitions), and
4
Bi = Â b W n (4.25) ji ji j
j = i +1
Pf = Â Âh e , ij
b ij Wij nis hn ij (4.27)
i =1 j =0
Dn3s ª I2 (4.30)
Ê ˆ Ê hn g W10 ˆ
2 2
2
hn e Á
Ë
 b3i W3i ˜ Á 2 +
¯ Ë s g I1 ˜¯
i =0
ab < a0 = (4.33)
Ê ˆ Ê hn g W10 ˆ
2 2
2
hn e Á
Ë
 b3i W3i ˜ Á 2 +
¯ Ë s g I1 ˜¯
i =0
È n(T ) ˘
2
p11 (T )
Q u (T ) = = Í n(T ) ˙ (4.36b)
p11 (T0 ) Î 0 ˚
where T0 = 300 K. Using spectrum data found in Ref. [21], the
temperature-dependent coefficients are plotted in Fig. 4.8. It can be
seen that Θu and Θg decrease with the decreasing of temperature.
When temperature drops to less than 200 K, ETU may become no
longer problematic to the cooling system since the ETU parameter is
tiny. When the temperature approaches 100 K, the small GSA in the
anti-Stokes domain will limit efficient cooling to lower temperatures.
On the contrary, Θe increases with the decreasing of temperature
because the ESA we adopted here is in the Stokes domain. By using
Eqs. 4.3, 4.23, and 4.33, the temperature-dependent α0 is plotted in
Fig. 4.9. It can be seen that when the sample temperature drops to
around 210 K, α0 = αb, which means no cooling enhancement occurs.
At temperatures lower than 210 K, α0 < αb, and the co-pumping
scheme will introduce extra heat to the cooling system (see Eq.
4.32). Therefore, we conclude that the co-pumping for cooling
enhancement only works at relatively high temperatures (for the case
discussed here, T > 210 K). Although it may not be helpful for cooling
at cryogenic temperatures, the co-pumping scheme can shorten the
time needed to reach the minimum attainable temperature of a
Ho3+ cooling system, due to the improved cooling performance at
temperatures of 210–300 K. To estimate the cooling improvement,
next we simulate the cooling power density and cooling efficiency of
the scheme at 300 K.
Generally, the cooling efficiency of an optical refrigeration system
dwindles when the input laser power is increased. As the pump
intensity approaches the saturation of the resonant absorption,
the parasitic absorption will become dominant, and eventually,
the cooling power density is lowered with the declining cooling
Efficient and Enhanced Holmium Optical Refrigeration via Co-Pumping 131
efficiency. To obtain the best cooling effect, the intensity of the pump
laser should be deliberately selected. In Ho3+ optical refrigeration
via the co-pumping scheme, the optimal pump intensities of GSA
and ESA for achieving the MCPD appear in pair. Figure 4.10a shows
the attainable MCPD in the Ho3+-doped YLiF4 crystal of various
dopant concentrations at 300 K. The optimal pump intensities for
reaching the MCPD are shown in Fig. 4.10b, and the corresponding
cooling efficiency is shown in Fig. 4.10c. It can be seen that an
MCPD of 7.5 × 107 W∙m–3 could be achieved in the 1.2% Ho3+:YLiF4
crystal. The MCPD in the co-pumping scheme is almost 3 orders of
magnitude higher than that of the conventional laser cooling with
Ho3+:YLiF4 crystals (see Section 4.2). The corresponding cooling
efficiency is about 4.2%, which is several times higher than what
could be obtained via the conventional pumping scheme. As it has
been mentioned, the enhancing effects are due to the cut-down in
the population density of the 5I5 manifold with an ESA of 5I5 Æ 5F5.
Since the decay from the 5F5 manifold is predominantly radiative,
the upconversion quenching is inhibited. Therefore, the dopant
concentration, as well as the pump intensity, could be increased,
resulting in an immense cooling enhancement.
Figure 4.10 At 300 K, (a) the maximum cooling power density (MCPD) as
a function of dopant concentration and (b) the corresponding
optimum pump intensities (OPIs). The solid curve is for GSA,
and the dashed curve is for ESA. (c) The corresponding cooling
efficiency. The background absorption is assumed to be 4.0 ×
10–4 cm–1.
Energy Transfer–Enhanced Laser Cooling in Rare Earth–Co-Doped Solids 133
pump laser is red-edge- tuned for Ho3+ ion excitation, the absorption
of Tm3+ ions is negligible (see Fig. 4.2d). For each excitation and de-
excitation cycle of Tm3+ ions, an extra thermal vibration energy of hν2
– hν1 will be released. According to ET theory, the density population
dynamic of the electronic energy levels of the above Ho-Tm cooling
system can be described by the following equations:
dn1 Pr
= - he,1W1,r n1 - W1,nr n1 - p13 n1 n3 + p02 n0 n2 (4.37a)
dt hn
dn2
= -he,2W2 ,r n2 - W2 ,nr n2 + p13 n1 n3 - p02 n0 n2 (4.37b)
dt
n0 + n1 = CHoN (4.37c)
n2 + n3 = CTmN (4.37d)
where ni (i = 0, 1, 2, 3) is the population of the ith level, as shown
in Fig. 4.12. Pr is the resonantly absorbed power density at pump
frequency ν. Wj,r and Wj,nr are the radiative decay rate and the
nonradiative decay rate, respectively (j = 1 for the Ho3+ 5I7 Æ 5I8
transition and j = 2 for the Tm3+ 3F4 Æ 3H6 transition). p13 and p02
are the temperature-dependent macroscopic ET rates [33]. CHo
and CTm are the percentage concentrations of Ho3+ and Tm3+ ions,
respectively.
where W1(2) is the total effective decay rate that includes both the
radiative decay rate with consideration of florescence reabsorption
and nonradiative decay rate. Substituting the analytical results of ni,s
into Eq. 4.39, a symmetrical expression is obtained,
Ê hext,1n 1 + hext,2n 2 hext,1n 1 - hext,2n 2 ˆ
hc = habs Á + ˜¯ - 1 (4.40)
Ë 2n 2n c
where ηext,1(2) is the external quantum efficiency for the Ho3+ 5I7 Æ
5I (Tm3+ 3F Æ 3H ) transition, and
8 4 6
vc = Pr/hA (4.41)
with
È 4 p13C TmW2 N ˆ ˘
1/2
Ê
2
Pr Pr
A=B-Í +ÁB - ˜ ˙ (4.42)
ÍÎ p02 - p13 hn Ë hn ¯ ˙˚
and
( p13C TmW2 + p02C HoW1 )N + W1W2
B= (4.43)
p02 - p13
Tm3+ ion equals zero, that is, the Ho3+–Tm3+ ET processes do not
exist (p13 = 0), the value of the characteristic frequency then equals
that of the pumping frequency (νc = ν). Equation 4.40 can be reduced
to
n1
hc n c =n
= hext,1habs - 1 (4.44)
n
This is exactly the cooling efficiency formula of the conventional
ASFC system. As a general expression of cooling efficiency for the
Ho-Tm ETLC system, Eq. 4.40 shows that the larger the characteristic
frequency of an ETLC system is, the larger the corresponding cooling
efficiency could be. The above analysis shows that the ETLC cooling
scheme is more efficient than the conventional ASFC scheme in
principle, so the right-hand side of Eq. 4.40 should be larger than
that of Eq. 4.44, indicating that νc > ν. Indeed, for the 0.3% Ho3+, 2%
Tm3+:YLiF4 crystal sample, the calculated result on the characteristic
frequency (with transition rates and macroscopic ET rates near 300
K found in Refs. [23, 34]) shows that the value of the characteristic
frequency is on the order of 2ν.
Next, we predict the cooling performance of an ETLC system
on the basis of spectrum data and material parameters of Ho3+,
Tm3+:YLiF4 crystals. The energy gap of Tm3+ 3F4–3H6 is 5810 cm–1
[23]. By using Eq. 4.20, the nonradiative decay rate of the Tm3+
3F Æ 3H transition in the YLiF crystal is calculated as 0.10 s–1.
4 6 4
The radiative decay rate of this transition is 66.67 s–1 [23]. The
external quantum efficiencies of Ho3+ 5I7 Æ 5I8 and Tm3+ 3F4 Æ 3H6
transitions in the YLiF4 host are calculated to be 0.9949 and 0.9985,
respectively. The average fluorescence wavelength for the Tm3+ 3F4
Æ 3H6 transition is found to be 1807 nm (see Fig. 4.2d).
Figure 4.13a shows the cooling efficiencies of the 0.3% Ho, 2%
Tm YLiF4 and the 0.3% Ho YLiF4 crystal samples as functions of
resonant absorption at a pump wavelength of 2070 nm. To point
out, since the dopant concentration of Ho3+ ions is relatively low, for
the sake of brevity, the model M1 (see Section 4.2) has been applied
in simulating Ho ASFC. As the resonant absorption increases,
the absorption efficiency will decrease; as a result, the cooling
efficiency of Ho ASFC drops. On the contrary, the cooling efficiency
of Ho-Tm ETLC first slightly rises with the increasing of the resonant
absorption, which is beneficial for boosting the cooling power density.
138 Laser Cooling of Solids around 2.07 Microns
In essence, as more Ho3+ ions are pumped to the excited state with
the increased absorption, the average distance between an excited-
state Ho3+ ion and a ground-state Tm3+ ion shrinks. Consequently,
the Ho3+–Tm3+ ET is accelerated and more Tm3+ ions participate in
the cooling cycle. At a resonant absorption of around 0.8 × 108 W/
m3, the 0.3% Ho, 2% Tm ETLC system reaches its maximum cooling
efficiency (5.08%). However, the cooling efficiency for the 0.3% Ho
ASFC system at a resonant absorption of 0.8 × 108 W/m3 is only
about 0.93%. The improving in the cooling efficiency by applying the
ET mechanism to a solid-state cooling system is quite remarkable.
Moreover, the maximum resonant absorption for cooling (ηc > 0) of
the 0.3% Ho, 2% Tm ETLC system is roughly two times of that of the
0.3% Ho ASFC system, indicating that the co-doped cooling scheme
could profoundly enhance the cooling power density. It can be seen
from Fig. 4.13b, at a resonant absorption of about 0.7 × 108 W/m3,
the MCPD for the 0.3% Ho ASFC system is 0.8 × 106 W/m3, whereas
at a resonant absorption of about 1.8 × 108 W/m3, the MCPD for the
0.3% Ho, 2% Tm ETLC system is 8 × 106 W/m3.
Figure 4.13 The cooling efficiencies (a) and the cooling power densities (b)
of Ho-Tm ETLC and Ho ASFC in the YLiF4 crystal as functions
of resonant absorption near 300 K. For the 0.3% Ho, 2% Tm:
YLiF4 crystal sample, the ET parameters are estimated as p13
= 6 × 10–24 cm3∙μs–1 and p02 = 8 × 10–23 cm3∙μs–1 [34]. The
background absorption is assumed to be 4.0 × 10–4 cm–1.
than 5% of the Ho3+ ions are pumped to the excited state for the
sample mentioned above. B that is defined by Eq. 4.43 will be more
than 20 times larger than Pr/hν. The right-hand side of Eq. 4.42 can
then be expanded as a power series of Pr/hν. Neglecting the terms
that are higher than second order, Eq. 4.42 can be approximately
expressed as
Pr p02C HoW1 - p13C TmW2
Aª (4.45)
hn p02C HoW1 + p13C TmW2
note that the host material should enable ASFC of each dopant in
the first place, and the doping concentrations should be restricted
within relatively low levels so that the deleterious ETU processes are
averted. Another couple of dopants for ETLC could be Er3+ and Yb3+,
since the Er3+ 4I11/2 and Yb3+ 2I5/2 states are very close in energy.
However, the drawback of Er-Yb ETLC is that the intermediate Er3+
11/2 Æ I13/2 transition usually produces a large amount of heat that
4I 4
4.5 Conclusions
This chapter theoretically investigates optical refrigeration in
Ho3+:YLiF4 and Ho3+, Tm3+:YLiF4 crystals. Due to the small energy
mismatch between transitions 5I7 Æ 5F8 and 5I7 Æ 5I5, ETU occurs
when the 5I5 Ho3+ ion is pumped in the anti-Stokes domain. Therefore,
a general cooling model with consideration of upconversion is
established to precisely describe laser cooling of Ho3+-doped
fluoride crystals. Meanwhile, a guide on selecting a host material for
optical refrigeration with multimanifold rare earth ions is provided.
Simulation results show that the multiphonon transitions from the
second and third excited states of Ho3+ ions block effective cooling.
To this end, a co-pumping scheme is presented to solve the ETU-
induced fluorescence quenching issue. An extra laser is added to the
cooling system to pump 5I5 Ho3+ ions to a higher excited state (5F5).
The beauty of applying the co-pumping scheme is that the 5F5 Ho3+
ions scarcely transit to the lower nearby state and mainly de-excite
radiatively. So, the large amount of heat produced by nonradiative
5I Æ 5I and 5I Æ 5I transitions in the conventional anti-Stokes
5 6 6 7
cooling case can be avoided, and both cooling efficiency and cooling
power can be prompted. Besides, there is another way of improving
the cooling performance of Ho3+ optical refrigeration: co-doping
Tm3+ and Ho3+ ions in the host. The Ho3+ and Tm3+ ions share a
similar electronic energy level structure, and their first excited
states are close in energy. As a result, ET from Ho3+ to Tm3+ ions
arises when Ho3+, Tm3+-co-doped fluoride crystals are pumped at
142 Laser Cooling of Solids around 2.07 Microns
around 2.07 μm. Since the energy gap of Tm3+ 3F4 - 3H6 is wider than
that of Ho3+ 5I7 - 5I8, host phonons will be annihilated when ET from
Ho3+ to Tm3+ ions takes place, bringing enhancement to the cooling
effect. However, one weak point of ETLC is that both of the two
dopants should be restricted within relatively low concentration
levels to avoid ETU-induced heating.
To conclude, ETU-induced fluorescence quenching is the main
obstacle to achieve net cooling in Ho3+-doped solids. Nevertheless,
by adopting the modified cooling schemes and the optimization
procedures presented in this chapter, it is possible to overcome
the ETU-induced heating and realize efficient optical refrigeration
around 2.07 μm.
Acknowledgments
The authors acknowledge financial support of the Specialized
Research Fund for the Program for New Century Excellent Talents
in University (grant no. NCET-11-269), the National Natural Science
Foundation of China (grant nos. 11204048, 61275138, 10804022,
60878016, and 61378027), the Fundamental Research Funds for
the Central Universities, the Special Financial Grant from the China
Postdoctoral Science Foundation (grant no. 2013T60345), and the
111 project to the Harbin Engineering University (grant no. B13015).
Appendix
For simplicity, the pump intensities for both GSA and ESA are
assumed to be low, so the population densities of 5I6, 5I5, and 5F5
manifolds are far less than the 5I7 manifold. In the conventional
cooling case, the steady-state population density (nis1, i = 0, 1, 2, 3)
of each manifold can be obtained from the following equations:
-( b30W30 + b31W31 + b32W32 )n3s1 + p11 (C )n1s1 = 0 (A.1a)
2
I1
s g (n0 s 1 - n1 s 1 ) - W10 n1 s 1 ª 0 (A.1b)
hn g
2 2
p11 C N
n3s1 ª (A.2)
Ê ˆ Ê hn g W10 ˆ
2 2
ÁË Â b3i W3i ˜ Á 2 +
¯Ë s I ˜¯
i =0 g 1
I2 Ê 2
ˆ
-s e n3s2
hn e
-Á
Ë
Âb 3i
W3i ˜ n3s2 + b 43W43 n4s2 + p11 n1s2 ª 0 (A.4)
¯
2
i =0
Í(1 - h43 ) +
hn e ÁË i = 0
b3i W3i ˜ ˙ Á 2 +
¯ ˚Ë
Â
s g I1 ˜¯
Î
The population density change of the 5I5 manifold induced by
adding the laser for ESA to the system is
-2
Ê hn g W10 ˆ
Dn3s = n3s2 - n3s1 = p11 C N ÁË 2 + s I ˜¯
2 2
g 1
(A.6)
ÏÔ È s e I2
2
˘
-1
Ê 2
ˆ ¸Ô
-1
¥ Ì Í(1 - h43 )
hn e
+ Â b3i W3i ˙ -Á
Ë
 b3i W3i ˜ ˝
¯ Ô˛
ÔÓ Î i =0 ˚ i =0
References
16. Condon, N. J., Bowman, S. R., O’Connor, S., Quimby, R. S., Mungan, C.
E. (2009). Optical cooling in Er3+:KPb2Cl5, Opt. Express, 17, pp. 5446–
5472.
17. Bowman, S. R., O’Connor, S., Condon, N. J., Friebele, E. J., Kim, W., Shaw,
B., Quimby, R. S. (2013). Non-radiative decay of holmium-doped laser
materials, Proc. SPIE, 8638, p. 863803.
18. Hehlen, M. P. (2009). Novel materials for laser refrigeration, Proc. SPIE,
7228, p. 72280E.
19. Dong, G. Z., Zhang, X. L. (2015). Role of upconversion in optical
refrigeration: a theoretical study of laser cooling with Ho3+ doped
fluoride crystals: erratum, J. Opt. Soc. Am. B, 32, pp. 2046–2047.
20. Walsh, B. M., Barnes, N. P., Bartolo, B. D. (1998). Branching ratios, cross
sections, and radiative lifetimes of rare earth ions in solids: application
to Tm3+ and Ho3+ ions in LiYF4, J. Appl. Phys., 83, pp. 2772–2787.
21. Walsh, B. M., Grew, G. W., Barnes, N. P. (2005). Energy levels and
intensity parameters of Ho3+ ions in GdLiF4, YLiF4 and LuLiF4, J. Phys.:
Condens. Mater., 17, pp. 7643–7665.
22. Barnes, N. P., Walsh, B. M. (2003). Ho:Ho upconversion: applications to
Ho lasers, J. Opt. Soc. Am. B, 20, pp. 1212–1219.
23. Walsh, B. M., Barnes, N. P., Petros, M., Yu, J., Singh, U. N. (2004).
Spectroscopy and modeling of solid state lanthanide lasers: application
to trivalent Tm3+ and Ho3+ in YLiF4 and LuLiF4, J. Appl. Phys., 95, pp.
3255–3271.
24. Hoyt, C. W., Hasselbeck, M. P., Sheik-Bahae, M., Epstein, R. I., Greenfield,
S., Thiede, J., Distel, J., Valencia, J. (2003). Advances in laser cooling of
thulium-doped glass, J. Opt. Soc. Am. B, 20, pp. 1066–1074.
25. Dexter, D. L. (1952). A theory of sensitized luminescence in solids, J.
Chem. Phys., 21, pp. 836–850.
26. Rapaport, A., Milliez, J., Bass, M., Cassanho, A., Jenssen, H. (2005).
Numerical model of the temperature dependence of the up-conversion
efficiency of fluoride crystals codoped with ytterbium and thulium, J.
Appl. Phys., 97, p. 123527.
27. van Dijk, J. M. F., Schuurmans, M. F. H. (1983). On the nonradiative and
radiative decay rates and a modified exponential energy gap law for
4f-4f transitions in rare-earth ions, J. Chem. Phys., 78, pp. 5317–5323.
28. Powell, R. C. (1998). Physics of Solid-State Laser Materials (Springer-
Verlag, New York).
29. Lee, T. H., Heo, J. (2005). 1.6 μm emission and gain properties of Ho3+
in selenide and chalcohalide glasses, J. Appl. Phys., 98, p. 113510.
146 Laser Cooling of Solids around 2.07 Microns
30. Payne, S. A., Chase, L. L., Smith, L. K., Kway, W. L., Krupke, W. F. (1992).
Infrared cross section measurements for crystals doped with Er3+,
Tm3+, and Ho3+, IEEE J. Quantum Electron., 28, pp. 2619–2630.
31. Librantz, A. F. H., Jackson, S. D., Gomes, L., Ribeiro, S. J. L., Messaddeq,
Y. (2008). Pump excited state absorption in holmium-doped fluoride
glass, J. Appl. Phys., 103, p. 023105.
32. Faiconieri, M., Salvetti, G. (1994). Effects of co-dopant concentrations
and excitation conditions on the 2 μm fluorescence dynamics in Tm,
Ho: YLF crystals. Appl. Phys. A, 59, pp. 253–258.
33. Walsh, B. M., Barnes, N. P., Bartolo, B. D. (1997). On the distribution of
energy between the Tm 3F4 and Ho 5I7 manifolds in Tm-sensitized Ho
luminescence, J. Lumin., 75, pp. 89–98.
34. Walsh, B. M., Barnes, N. P., Bartolo, B. D. (2000). The temperature
dependence of energy transfer between the Tm 3F4 and Ho 5I7
manifolds of Tm-sensitized Ho luminescence in YAG and YLF, J. Lumin.,
90, pp. 39–48.
Chapter 5
Steven R. Bowman
Laser Physics Branch, Naval Research Laboratory, Washington, DC 20375-5320, USA
steven.bowman@nrl.navy.mil
Since the invention of the laser, engineers have searched for new
ways to increase the laser’s power and brightness. These efforts
have largely focused on improved ways to remove the waste heat
that is generated in the lasing medium. This chapter reviews a novel
approach to the problem of laser power scaling: minimizing the
waste heat through the incorporation of anti-Stokes fluorescence
cooling. This approach has been shown to significantly reduce
heat generation in carefully controlled laser systems. This chapter
reviews the theory and practice of this novel mode of laser operation.
5.1 Introduction
One of the more useful properties of lasers is their ability to generate
extremely high power densities. Take, for example, a commercial
fiber laser operating continuously with an average power of 1 kW.
Inside the small core of the fiber the beam propagates at intensities
as high as 1 GW/cm2. For comparison, this intensity is 10 billion
times higher than terrestrial sunshine. Even the hardest metals
are instantly vaporized when placed near the focus of this beam.
Yet the laser that generates these intensities is expected to operate
reliably for years. The challenge for the laser designers is to cleanly
manipulate such high intensities without damaging the laser
medium.
What happens to a material when it is exposed to a high-intensity
laser depends principally on the amount of power absorbed. Most
materials convert absorbed laser power directly into heat. When the
absorption is described by a Beer’s law coefficient of α and the laser
intensity is IL, then the absorbed power density is simply αIL. So any
material with significant absorption placed in the beam described
above would see GW/cm3 heat loads and be instantly vaporized.
Even the most transparent materials, such as a pure silica fiber,
have weak background absorption with coefficients of order of 10–6
cm–1. So propagation of the 1 kW laser beam described above will
still generate 1 kW/cm3 of waste heat. The core of an optical fiber
survives this extreme heat loading by spreading the thermal load
over a large surface area. Less transparent materials, such as a laser
gain medium, can see even higher thermal loads. As a result, thermal
loading remains the primary obstacle to scaling up the average
power of most laser systems.
irradiance IL. For simplicity, we assume that the pump and laser fields
are directional, monochromatic, and polarization independent. This
is not always the case, but the model is easily modified to include
broader intensities and nonisotropic media. Our derivation also
assumes that ionic dephasing is very rapid compared to tR, allowing
us to consider only incoherent excitation. With these definitions and
mass conservation, we can write the local rate equation as
dN2 IPlPs A (lP ,T )
= ◊ ÈÎ NT - N2 b(lP ,T )˘˚
dt hc
I l s ( l ,T )
+ L L A L ◊ ÈÎ NT - N2 b(lL ,T )˘˚ - N2 t R .(5.2)
hc
The first term on the right describes the pump rate, the second
term describes the laser rate, and the last term describes the total
spontaneous decay rate of the upper manifold. Applying energy
conservation to this laser system, we recognize that the difference
between the absorbed power and the emitted power must be the
local heat generation rate, H. With the definitions above the power
density of waste heat will be
hc N2
H = IPs P ◊ È NT - ˘+I ◊ È NT - ˘-
N2 N2
Ls L . (5.3)
ÎÍ b P ˚˙ ÎÍ b L ˚˙ lF t R
lF ∫
Ú band
l ◊ IF ( l ) d l
. (5.4)
Ú
band
IF ( l ) d l
1
=
Z1 T( ) exp Ê e2 ˆ Èexp Ê -hc ˆ - exp Ê -hc ˆ ˘ , (5.8)
bPL Z2 T( ) ÁË kT ˜¯ ÍÎ ÁË lL kT ˜¯ ÁË l kT ˜¯ ˙
P ˚
Ú s A (l , T ) exp( - hc l kT ) dl l
4
and
d lF = band
.
Ú s A (l , T ) exp( - hc l kT ) dl l
5
band
H=
[
NT hc lF ( lL - lP )iP iL bP bL + lL ( lF - lP )bP bPL iP - lP ( lL - lF )bL bPL iL ].
lF t R [ ]
l P l L 1 + iP + iL b P L
(5.12)
Unlike the radiative power densities, heat builds up in the lasing
medium, leading to temperature gradients and detrimental thermo-
optic distortions. For a steady-state laser, temperature gradients
must build until equilibrium is established. In a high-power laser
system, significant additional power and hardware will be required
to manage the thermal effects. It is therefore reasonable to reduce
heating, while maintaining laser performance. In the next section we
will examine laser optimization strategies using this simple model.
We illustrate the above relations using the radiative properties
of Yb:YAG. Figure 5.1 shows the energy structure and cross-sectional
spectra at room temperature. The mean fluorescence wavelength and
radiative lifetime of Yb:YAG are reported to be 1010 nm and 0.979
ms, respectively, at room temperature [5, 10]. For these calculations
the laser wavelength will be chosen at 1050 nm, which corresponds
to the lowest-energy transition from the 2F5/2 level of Yb:YAG. The
peak absorption cross section of the 1050 nm line is only 6.0E-23
cm2, leading to a saturation intensity of 70 kW/cm2.
In Fig. 5.2 we illustrate how the cross-sectional parameters
defined above vary with the choice of operating conditions in the
Yb:YAG system. It shows that the βs can vary by many orders of
magnitude for reasonable choices of operating wavelengths and
temperature. The single-wavelength parameter b(l, T) is plotted
using lines with triangles for 100 K, 200 K, and 300 K. At short
wavelengths, b(l, T) approaches unity for all temperatures, but it
falls off rapidly with wavelength and temperature. Since b(l, T) is the
156 Optically Cooled Lasers
-20 10327
10
λP λL
-21
10
-22 λP λL 785
10
2F 612
7/2 565
0 cm-1
880 920 960 1000 1040 1080 Yb:YAG
Wavelength (nm)
Figure 5.1 (a) Room-temperature absorption (lP) and emission (lL) cross
sections measured in optically thin samples of Yb:YAG. (b)
Yb:YAG energy levels at room temperature.
10
0.1
0.01
0.001
0.0001
0.00001
0.9 0.92 0.94 0.96 0.98 1
lP/lL
Figure 5.2 Plot of the fractional cross sections discussed in the text
evaluated for Yb:YAG. Lines with triangles delineate βP. Lines
with circles are βPL/βP and solid lines are βPL/βL. Values are
calculated for several temperatures, assuming λL = 1050 nm:
300 K (red), 200 K (green), and 100 K (blue).
Theory of Low–Quantum Defect Laser Materials 157
hm ∫
L
=
[
lP 1 - bPL bP iP ] . (5.13)
P [
lL 1 + bPL bL iL ]
Note that the dependence of the efficiency on intensity has
shifted from that in Eq. 5.10. Now increasing either the pump or the
laser intensity increases hm.
We should be careful to distinguish the laser material efficiency,
hm, from the more commonly used laser optical efficiency. The laser
optical efficiency will be proportional to hm but will also depend
158 Optically Cooled Lasers
Xm ∫
H
=
[lF (lL - lP ) iPiL + lL (lF - lP ) iP bPL bL - lP (lL - lF ) iL bPL bP ].
L lF lPiL [iP - bPL bP ]
(5.14)
Recognizing that the heat generation can be positive or negative,
we define Xm in terms of the absolute magnitude. Note that Xm is
defined in terms of laser power and not absorbed pump power. The
heating parameter more commonly used in the literature would be
the product of hm and Xm. The definitions used here will lead more
directly to an optimization procedure.
The above equations describe the performance of a purely
radiative cw laser. Laser power can theoretically be optimized by
either maximizing hm or minimizing Xm. This would be accomplished
through the appropriate selection of the operating intensities and
wavelengths. Note that the heating fraction depends strongly on
wavelength separations, while the laser efficiency is independent of
the fluorescence wavelength.
Consider first the approach of improving laser performance
by maximizing the laser material efficiency. Equation 5.13 reveals
that whenever the laser material is pumped far above inversion, the
material efficiency will approach lP/lL. Lower intensities reduce the
material efficiency and longer pump wavelength generally increase
it. However, we can see from Fig. 5.2 that as lP approaches lL, bPL/
bP and bPL/bL increase rapidly, raising the laser’s threshold and
decreasing efficiency. This optimization approach is illustrated in
Fig. 5.3 for the idealized Yb:YAG lasing at 1050 nm. As expected, laser
material efficiency initially rises linearly with pump wavelength and
then falls to zero as lP approaches lL. Good efficiencies require that
intensities exceed the saturation value. At the highest intensities
considered, the room-temperature material efficiency reaches 88%
when the pump wavelength is 1005 nm. Cooling is seen to improve
material efficiency somewhat at longer wavelengths. For the highest
Theory of Low–Quantum Defect Laser Materials 159
1.0
0.8
0.6
ηm
0.4
i = 16
= 8
= 4
0.2 = 2
= 1
= 0.5
0.0
0.90 0.92 0.94 0.96 0.98 1.00
λP / λL
Figure 5.3 Plot of the material efficiency for an ideal Yb:YAG laser with
λL = 1050 nm and several values of the scaled intensity. For
the purpose of this illustration, the scaled pump and laser
intensities are set equal. Solid curves are for 300 K and dashed
curves are for 100 K.
1
λF
0.1
0.01
m
Ξ
i = 16
= 8
= 4
0.001 = 2
= 1
= 0.5
0.0001
0.9 0.92 0.94 0.96 0.98 1
λP / λL
Figure 5.4 Plot of laser material heating parameter for idealized Yb:YAG.
Lasing is at 1050 nm and the temperature is 300 K. The vertical
line marks the position of the mean fluorescence wavelength.
As before, the scaled pump and laser intensities are set equal.
Theory of Low–Quantum Defect Laser Materials 161
1 0.05
0.8 0.04
Quantum Defect
0.6 0.03
(lP – lF)/lP
ηRBL
0.4 0.02
(lL – lP)/lL
0.2 0.01
0 0
3
0.1 1 10 100 1×10
iPmin
Figure 5.5 The solid curve is the predicted RBL material efficiency as a
function of the minimum pump intensity. The dashed lines
give the laser quantum defect (lL – lP)/lL and the fluorescence
quantum defect (lP – lF)/lP. This calculation assumes the
conditions of the Yb:YAG laser discussed in the text.
162 Optically Cooled Lasers
The dashed curves in Fig. 5.5 illustrate how the QDs for the laser
and fluorescence emission vary with minimum pump intensity.
When the QDs are equal the heat generated by one laser photon
is on average offset by the cooling from one fluorescent photon.
At this point the RBL efficiency is 50%. As the pump wavelength is
increased, laser QD heating decreases and fewer fluorescent photons
are required to maintain balance. This requires the laser efficiency
increases to maintain the RBL condition. For example, at iPmin = 10,
a single fluorescent photon should balance the defect of three laser
photons.
Comparing Figs. 5.3 and 5.4 reveals that optimums for the two
approaches do not occur at the same pump wavelength. The RBL
points occur at longer wavelengths for the same intensity. Taking
the iP = iL = 2 case, for example, an ideal 1050 nm Yb:YAG laser
pumped at 970 nm is predicted to have hm = 0.593 and Xm = 0.102.
The same laser pumped at 1026 nm would see hm = 0.402, with
no heat generation. At the highest intensities plotted, iP = iL = 16,
the maximum possible efficiency is 0.882, while the RBL efficiency
is 0.745. Since thermal loading requires active cooling to maintain
temperature and beam quality, these systems’ overall efficiencies
are likely to be comparable.
A third approach for improving laser performance is motivated
by the special case of no lasing. When laser intensities are small, the
laser material heating efficiency simplifies to become proportional
to lP – lF. So if the laser medium is pumped at its mean fluorescent
wavelength, there will be no heat generated by the pump. This
approach might be indicated when the stored energy is not fully
extracted from the laser medium, as often occurs in bulk laser
geometries or during pulsed mode operation. For a laser operating
with the condition of lP = lF, the fractional laser heating simplifies
to the laser’s QD:
Xm = (1 – lL/lF). (5.17)
Note that this is the only pump-intensity-independent heat-
loading solution for an LQDL. Heat loading would be simply
proportional to laser power. In the example plotted in Fig. 5.3 this
operating point is labeled by a vertical dashed line. Pumping at 1010
nm and lasing at 1050 nm give a heating parameter of 3.8%. Blue-
shifting the laser wavelength could readily reduce the laser heating
parameter to below 1%, while maintaining good efficiency.
Theory of Low–Quantum Defect Laser Materials 163
1.0
0.8
0.6
η´m
0.4
i = 16
= 8
= 4
0.2 = 2
= 1
= 0.5
0.0
0.90 0.92 0.94 0.96 0.98 1.00
λP / λL
Figure 5.6 Plot of the predicted laser material efficiency for the same
conditions used in Fig. 5.3. Material parameters have been
modified, as described in the text.
¢ ∫
Xm
H¢ t
= F
[lF¢ (lL - lP ) iPiL bP bPL + lL ( lF¢ - lP )iP bP bL - lP ( lL - lF¢ )iL ]
L¢ t F¢ [
lF¢ lPiL iP bP bPL - 1 ]
l Ê tR n t ˆ È i b i b + 1 ˘ È iL bLda L + iP bPda P lL lP ˘
+ L ÁË t-1+ 2 R ˜ Í P P L L ˙+Í ˙.
lP F t Q¢ ¯ Î i P bP bPL - 1 ˚ Î n2 - bL ˚
(5.23)
Here the first term on the right accounts for radiative trapping.
The second term accounts for quenching and the third accounts for
background absorption. Here the approximation, n¢2 ≈ n2, found to
be valid in Eq. 5.21 has been used. Figure 5.7 plots X¢m for the Yb:YAG
example for the same conditions used in Fig. 5.6. Comparing this plot
with that of the idealized Yb:YAG in Fig. 5.4 reveals the impact of
adding realistic material losses. Linear and quadratic quenching red-
shift the crossover wavelength by 5 nm and background absorption
adds another 1 nm shift. When radiative trapping is added to the
model t ¢F = 1.25 ms, thermal loading decreases and the crossover
wavelength red shift increases to 15 nm. This narrows and disrupts
the RBL resonance at lower intensities. Background absorption
heating is stronger at higher intensities. For the highest intensities,
a fractional background absorption of daL = daP = 10–3 disrupts the
168 Optically Cooled Lasers
Figure 5.8 Edge-pumped Yb:KGW disk laser with a laser quantum defect
of 4.8%.
0.02
0
i = 16
Gain Coefficient (cm–1)
= 8.0
= 4.0
= 2.0
= 1.0
0.01
− 0.1 = 0.5
− 0.2 0
The pump beams were injected into the resonator off-axis using
a flat HR mirror with a 3 mm on-axis hole. The broadband HR back-
resonator mirror reflected the pump beams back to the center of
the rod. The beams crossed near the center of the rod, producing a
roughly 2 mm diameter mode volume. The concave output mirror
Low–Quantum Defect Laser Experiments 173
90
Laser Power (W)
Slope = 60%
Slope = 41%
80
70
60
50
280 300 320 340 360 380 400 420
Incident Pump Power (W)
Figure 5.11 Comparison of the laser performance of the Yb:YAG resonator
for two doping densities: 3.1E20 ions/cc (dots) and 8.1E20
ion/cc (squares).
For the above laser configuration the minimum heat load condition
was found to require an incident pump intensity of iP = 2.1 and an
on-axis intercavity laser intensity of iL = 1.0. This is a close to the
predicted value of the nonideal model developed in Section 5.3.3.
This is consistent with the strong radiation trapping observed in
these experiments. Fluorescence lifetimes from these laser rods were
found to be 2.0 ms, twice as long as the measured value τR. Near the
RBL resonance, the laser power was 80 W and optical efficiency was
174 Optically Cooled Lasers
26%. This is consistent with the nonideal LQDL model that predicts
the average material efficiency for the rod should be 27%.
Figure 5.13 shows a typical thermal image of the laser rod near
the RBL condition. For optimized conditions heat loading in the
laser rod was too low to measure with the thermal camera. This
corresponds to a material heat parameter X´ < 10–4. This was the
first truly RBL.
250
200
Power (W)
150
100
50
0
0 10 20 30 40
Time (s)
Figure 5.12 Power and near-field beam profile of the 3% Yb:YAG resonator.
Figure 5.13 A typical thermal image of the 3 × 120 mm Yb:YAG rod inside
the radiation-balanced laser. This image was captured after
10 seconds of lasing while being pumped with 372 W at 1030
nm. The false color scale corresponds to ±5°C, as shown on the
right. Fluorescence from the laser rod heats the mirror mount
on the left, but the thermally insulated laser rod remains very
close to the ambient temperature.
Low–Quantum Defect Laser Experiments 175
and the success of the 2-for-1 pumping scheme using 0.8 mm diodes.
Nevertheless, a resonant-pumped thulium laser system has recently
been reported with slope efficiencies of 90% [24]. This system used
0.8 mm diodes to pump a thulium fiber laser operating tuned to
1.908 mm. This fiber was then used to pump another thulium fiber
lasing at 2.005 mm, leading to a laser QD of 4.8%. As with the erbium
demo, fluorescent cooling of thulium was not possible with this
pump wavelength.
One of the best-studied midwave LQDL systems is based on
the 2.1 mm holmium transition. First demonstrated in 1992, this
laser is usually excited with a diode-pumped 2.0 mm thulium
laser, as discussed above [16, 18, 62]. This laser scheme provides
coherent beam combination of multiple thulium lasers and avoids
strong upconversion that occurs in concentrated Tm and Tm/
Ho laser materials [15]. The holmium LQDL scheme has received
considerable attention due to relatively eye-safe wavelength of 2.1
mm and the superior transmission through the atmosphere.
In 2015 Loiko et al. demonstrated record-high efficiency in a
holmium-doped KLu(WO4)2 microchip laser [47]. They achieved
84% slope efficiency at 2.105 mm pumping with a thulium laser at
1.945 mm at only 4.4 kW/cm2. Unfortunately optical efficiency was
low for this demo due to poor pump absorption and operation near
threshold.
A low-power experiment demonstrated 72% slope efficiency
from a 2080 nm holmium-doped ZBLAN fiber. The thulium fiber
pump source was operated at 2051 nm, leading to a very low laser
QD of 1.4% [35]. This LQDL scheme has the potential for optical
cooling since lF = 2010 nm in Ho:ZBLAN.
Other recent laser demonstrations have explored holmium
doped into silica fibers. Average powers up to 400 W have been
generated but with lower optical efficiencies [54, 37]. Current work
on this material seeks to reduce the hydroxyl-induced background
absorption and lower the multiphonon quenching in holmium-
doped fibers [30, 42].
5.5 Summary
LQDLs are important because they have the potential to operate
with both high efficiency and high power. Optical pumping near the
Summary 179
also demonstrated high efficiency but are not yet capable of scaling
to high powers.
5.6 Appendix
Radiation trapping significantly impacts the performance of LQDLs.
Here we will develop an approximate solution for the excitation
density in an optically thick laser medium [10]. Using the notation
developed in Section 5.3, the rate equation for the local excitation
density will now include a term for quadratic quenching and
reabsorption of the local fluorescence:
dN2 IPs P Ê N ˆ Is Ê N ˆ N
= Á NT - 2 ˜ + L L Á NT - 2 ˜ - 2
dt hn P Ë b P ¯ hn L Ë bL ¯ t F
(5.24)
N2 Ê I (l ) ◊ s A (l ) ˆ Ê N ˆ
- 2 + Á F ˜ Á NT - 2 ˜ .
NTt Q Ë hn F ¯Ë b (l ) ¯
Here IF(l) is the local fluorescence spectral irradiance at
wavelength l. Unlike the pump and laser fields, IF(l) will be
spectrally broad with complex directionality. The last term in the
equation is the rate of fluorescence reabsorption integrated over the
entire emission spectrum as well as all directions and polarizations.
For precise evaluation of N2, the local fluorescence intensity
should be the integrated contributions from the entire laser crystal
volume. Computing IF(l) from first principles requires a detailed
knowledge of the excitation state of the entire lasing medium, as well
as its surface reflectivity. This approach would require significantly
more computation time to build up a self-consistent steady-state
spatial solution for all three fields.
To simplify the problem, we can approximate the laser crystal as
an optical cavity in which tC is the average lifetime of the available
cavity modes. This lifetime might be computed from numerical
ray traces or treated as an empirical parameter derived from
fluorescence decay data. Fluorescence reabsorption reduces the
photon lifetime within the crystal cavity according to
1 1 c ◊ a(l )
= + . (5.25)
tP tC n
Here c is the speed of light, n is the index of refraction, and a(l)
is the absorption coefficient. For simplicity we will assume that tC
Summary 181
1 1
∫ ◊
Ú
g¢(l ) dl
and
t F¢ t F
Ú
g(l ) dl
(5.29)
a(l )
1 1 c tC Ú b( l )
g¢(l ) dl
∫ + ◊ .
t Q¢ t Q nt F
Ú g(l ) dl
182 Optically Cooled Lasers
A ∫ C 2 + 4 (iPbP + iL bL ) t Q¢ t R , C ∫ (iP + iL ) t R + 1 t F¢ ,
(5.32)
B ∫ ( A - C ) A + C and NSS = NT ◊ ( A - C ) ◊ t Q¢ 2.
References 183
References
1. Abate, J. A., Lund, L., Brown, D., Jacobs, S., Refermat, S., Kelly, J., Gavin, M.,
Waldbillig, J., Lewis, O. (1981). Active mirror: a large-aperture medium
repetition rate Nd:glass amplifier, Appl. Opt., 20(2), pp. 351–361.
2. Albrecht, G., Sutton, S., George, V., Sooy, W., Krupke, W. (1998). Solid
state heat capacity disk laser, Laser Part. Beams, 16, pp. 605–625.
3. Albrecht, G. F., Eggleston, J. M., Ewing, J. J. (1986). Design and
characterization of a high average power slab YAG laser, IEEE J.
Quantum Electron., QE-22, pp. 2099–2106.
4. Beach, R. J. (1996). CW theory of quasi-three level end-pumped laser
oscillators, Opt. Commun., 123, pp. 385–393.
5. Biswal, S., O’Connor, S. P., Bowman, S. R. (2006). Non-radiative losses in
Yb:KGd(WO4)2 and Yb:Y3Al5O12, Appl. Phys. Lett., 89(9), pp. 0919111–
0919113.
6. Bourdet, G. L. (2000). Theoretical investigation of quasi-three-level
longitudinally pumped continuous wave lasers, Appl. Opt., 39(6), pp.
966–971.
7. Bowman, S. R. (2015). Optimizing average power in low quantum
defect lasers, Appl. Opt., 54(31), pp. F78–F84.
8. Bowman, S., Ganem, R. J., Brown, C. G. (2014). Optical cooling in multi-
level systems, in Proceedings of the Laser Refrigeration of Solids VII,
SPIE 90000G, doi:10.1117/12.2040073.
9. Bowman, S. R., O’Connor, S., Condon, N. J., Friebele, E. J., Kim, W., Shaw,
B., Quimby, R. S. (2013). Non-radiative decay of holmium-doped laser
184 Optically Cooled Lasers
24. Creeden, D., Johnson, B. R., Setzler, S. D., Chicklis, E. P. (2014). Resonantly
pumped Tm-doped fiber laser with >90% slope efficiency, Opt. Lett.,
39, pp. 470–473.
25. Eggleston, J., Kane, T., Kuhn, K., Unternahrer, J., Byer, R. L. (1984). The
slab geometry laser 1. Theory, IEEE J. Quantum Electron., 20(3), pp.
289–301.
26. Fan, T. Y., Ripin, D. J., Aggarwal, R. L., Ochoa, J. R., Chann, B. Tilleman, M.,
Spitzberg, J. (2007). Cryogenic Yb3+-doped solid-state lasers, IEEE J.
Select. Top. Quantum Electron., 13(3), pp. 448–458.
27. Fan, T. Y. (1992). Optimizing the efficiency and stored energy in quasi-
three-level lasers, IEEE J. Quantum Electron., 28, pp. 2692–2697.
28. Fernandez, J., Garcia-Adeva, A. J., Balda, R. (2006). Anti-Stokes laser
cooling in bulk erbium-doped materials, Phys. Rev. Lett., 97, pp.
0330011–0330014.
29. Fowler, W. B., Dexter, D. L. (1962). Relation between absorption and
emission probabilities in luminescent centers in ionic solids, Phys. Rev.,
128(5), pp. 2154–2165.
30. Friebele, E. J. , Askins, C. G., Peele, J. R., Wright, B. M., Condon, N. J.,
O’Connor, S., Brown, C. G., Bowman, S. R. (2014). Ho-doped fiber
for high energy laser applications, Proc. SPIE, 8961, pp. 8961201–
8961209.
31. Gapontsev, D. (2008). Recent progress on high power fiber lasers, in
11th Annual Directed Energy Symposium, DEPS 1–27.
32. Garbuzova, D., Kudryashov, D. I., Dubinskii, M. (2005). 110 W, 0.9 J
pulsed power from resonantly diode-laser-pumped 1.6-μm Er:YAG
laser, Appl. Phys. Lett., 87, pp. 1211011–1211013.
33. Giesen, A., Speiser, J. (2007). Fifteen years of work on thin-disk lasers:
results and scaling laws, IEEE J. Sel. Top. Quantum Electron., 13(3), p.
598.
34. Giesen, A., Hugel, H., Voss, A., Witting, K., Brauch, U., Opower, H. (1994).
Scalable concept for diode-pumped high-power solid-state lasers,
Appl. Phys. B, 58, pp. 365–372.
35. Guhur, A., Jackson, S. D. (2010). Efficient holmium-doped fluoride
fiber laser emitting 2.1 μm and blue upconversion fluorescence upon
excitation at 2 μm, Opt. Express, 18(19), pp. 20164–20169.
36. Hehlen, M. P., Epstein, R. I., Inoue, H. (2007). Model of laser cooling
in the Yb3+-doped fluorozirconate glass ZBLAN, Phys. Rev. B, 75, pp.
144302–144313.
186 Optically Cooled Lasers
37. Hemming, A., Simakov, N., Davidson, A., Bennetts, S., Hughes, M.,
Carmody, N., Davies, P., Corena, L., Stepanov, P., Haub, J., Swain, R.,
Carter, A. (2013). Monolithic cladding pumped holmium-doped fibre
laser, IEEE Conference on Lasers and Electro-Optics, CW1M.1.
38. Hoyt, C. W., Sheik-Bahae, M., Epstein, R. I., Edwards, B. C., Anderson, J.
E. (2000). Observation of anti-Stokes fluorescence cooling in thulium-
doped glass, Phys. Rev. Lett., 85, pp. 3600–3603.
39. Injeyan, H., Goodno, R. (2011). High Power Laser Handbook (McGraw-
Hill).
40. Jauregui, C., Eidam, T., Limpert, J., Tunnermann, A. (2011). The
impact of modal interference on the beam quality of high-power fiber
amplifiers, Opt. Express, 19, pp. 3258–3271.
41. Kane, T. J., Eggleston, J. M., Byer, R. L. (1985). The slab geometry laser:
part II; thermal effects in a finite slab, IEEE J. Quantum Electron., QE-
21, pp. 1195–1210.
42. Kim, W., Bowman, S. R., Baker, C., Florea, C., Villalobos, G., Shaw, B.,
Sadowski, B., Hunt, M., Aggarwal, I., Sanghera, J. (2014). Holmium
doped laser materials for eye-safe solid state laser application, Proc.
SPIE, 90814, pp. 9081051–9081056.
43. Klopp, P., Petrov, V., Griebner, U. (2003). Potassium ytterbium tungstate
provides the smallest laser quantum defect, Jpn. J. Appl. Phys., 42, pp.
246–248.
44. Lacovara, P., Wang, C. A., Choi, H. K., Aggarwal, R. L., Fan, T. Y. (1991).
Room-temperature InGaAs diode-pumped Yb-YAG laser, OSA Proc. Adv.
Solid-State Lasers, 10, pp. 283–285.
45. Larat, C., Schwarz, M., Lallier, E., Durand, E. (2014). 120 mJ Q-switched
Er:YAG laser at 1645 nm, Opt. Express, 22(5), pp. 4861–4866.
46. Li, C., Liu, Q., Gong, M., Chen, G., Yan, P. (2004). Modeling of radiation-
balanced continuous-wave laser oscillators, J. Opt. Soc. Am. B, 21(3),
pp. 539–542.
47. Loiko, P., Maria Serres, J., Mateos, X., Yumashev, K., Kuleshov, N.,
Petrov, V., Griebner, U., Aguiló, M., Díaz, F. (2015). In-band-pumped
Ho:KLu(WO4)2 microchip laser with 84% slope efficiency, Opt. Lett.,
40(3), pp. 344–347.
48. Mandal, A., Klimek, D. E. (2010). Textron’s J-HPSSL 100 kW ThinZag,
IEEE Conference on Lasers and Electro-Optics, JThH2.
49. Marmo, J., Injeyan, H., Komine, H., McNaught, S., Machan, J., Sollee, J.
(2009). Joint high power solid state laser program advancements at
Northrop Grumman, Proc. SPIE, 7195, pp. 7195071–7195076.
References 187
64. Takaichi, K., Yagi, H., Shirakawa, A., Ueda, K., Hosokawa, S., Yanagitani,
T., Kaminskii, A. A. (2004). Lu2O3:Yb3+ceramics a novel gain material
for high-power solid-state lasers, Phys. Status Solidi A, 202, pp. R1–R3.
65. Ward, B., Robin, C., Dajani, I. (2012). Origin of thermal modal
instabilities in large mode area fiber amplifiers, Opt. Express, 20, pp.
11407–11422.
66. White, J. O., Dubinskii, M., Merkle, L. D., Kudryashov, I., Garbuzov, D.
(2007). Resonant pumping and upconversion in 1.6 μm Er3+ lasers, J.
Opt. Soc. Am. B, 24(9), pp. 2454–2460.
67. Wright, J. (1976). Up-conversion and excited state energy transfer in
rare-earth doped materials, Radiationless Processes in Molecules and
Condensed Phases, Vol. 15 of Topics in Applied Physics, Chapter 4, pp.
239–295.
68. Yao, T., Ji, J., Nilsson, J. (2014). Ultra-low quantum-defect heating in
ytterbium-doped aluminosilicate fibers, J. Lightwave Tech., 32(3), pp.
429–434.
Chapter 6
Stephen C. Rand
MURI Center for Dynamic Magneto-Optics, Appl. Physics & EECS Optics & Photonics
Laboratory, 3102 ERB1, 2200 Bonisteel Blvd., University of Michigan, MI, USA
scr@umich.edu
6.1 Introduction
The topic of laser cooling began with the articles of Pringsheim
[1] and Hansch [2] long before any laboratory experiments were
undertaken. These early papers proposed the use of optical radiation
to remove energy from matter for the purpose of refrigeration, but
the mere idea that light could transport thermal energy was widely
doubted until Landau’s thermodynamic analysis showed that
radiation fields had entropy that depended on the statistics and
mode structure of the field [3]. The influence of various states of the
field on solid state cooling was examined in detail later [4], but once
it was realized that entropy of fields could be exchanged for entropy
of matter, the stage was already set for a concerted worldwide effort
to alter the temperature of collections of atoms with light.
Experiments to observe laser cooling of atomic motion in
gases originated from investigations of the mechanical effects of
ν νfl
1
Figure 6.1 Anti-Stokes fluorescence emission in a two-level system.
q→
+ _
_
+ + +
_ _ T
+
Ω q→
_ +
_
+ T+ +
_ _
+ q
π /a
Figure 6.2 (Left) Dispersion curves illustrating the variation of phonon
frequency W with wavevector q for transverse acoustic (TA)
and transverse optic (TO) modes. The lattice constant is a.
(Right) Particle displacements in a diatomic linear lattice for
TO (upper) and TA (lower) modes of the same wavelength.
Here
- +
E1 ( z , t ) = xk (a 1 e - i(w1t - kz ) + a 1 ei(w1t - kz ) ) (6.11)
and
+ -
E2* ( z , t ) = xk + q (a 2 ei(w2t -[ k + q]z ) + a 2 e - i(w2t -[ k +q ]z ) ) (6.12)
When Eqs. 6.8 and 6.10–6.13 are combined with Eq. 6.9 the
interaction Hamiltonian becomes
H I = -( NM )-1/2 (e ◊ —V )Q(W , t )exp[iq ◊ R0 ]
- -
= f (b q ak++ q a k + h.c .) (6.14)
where the interaction strength is defined as f *₌ (2ħNMΩ)–1/2Vγqξkξk+q.
V is the volume of interaction. The conservation of momentum in the
interaction is illustrated in Fig. 6.3.
Ω, q
ω2 , k2
θ
ω1 , k1
H= ÂH
q ,l
ql = Â W
q ,l
+ -
ql ( bql bql + 12 ) (6.17)
+ -
Since n ql = b ql b ql is the number of phonons in mode q, the
summation in Eq. 6.17 describes the total vibrational energy of the
medium as a weighted sum of the individual mode energies which in
turn determines the sample temperature.
With the interaction Hamiltonian of Eq. 6.14 in hand,
optically driven dynamics of two optical waves interacting
with an acoustic mode can be analyzed. The static Hamiltonian
+ - + - + -
H = w a 1 a 1 + w a 2 a 2 + Wb b plays no role in steady-state
0 1 2
dynamics. For example, the change of phonon occupation with time
can be calculated using the Heisenberg equation of motion
.
n b = (i / )[H0 + Hint , n ]
- + -
{ + -
= (i / ) f (b a 2 a 1 + h.c .), b b
}
- + - + -
{ + - + - +
= if b a 2 a 1 (b b ) - b (b b )a 2 a 1 (6.18) }
The first term on the right determines the rate of annihilation
of phonons, while the second term is the rate of creation. We can
198 Methods for Laser Cooling of Solids
+
show this explicitly by solving for the temporal behavior of b and
-
b , again using the Heisenberg picture.
+ + - - + + + + - - +
(db /dt ) = iWb + if b a 1 a 2 b - if b a 1 a 2 b - G b b (6.19)
- - + + - - - - + + -
(db / dt ) = -iWb - if b a 1 a 2 b + if b a 1 a 2 b - G b b (6.20)
Making use of the slowly varying envelope approximation (SVEA),
steady-state solutions for the operator amplitudes are simply
+
{ - - + + + + - -
}
b = if b a 1 a 2 b - if b a 1 a 2 b / G b
-
{ + + - - - - + +
}
b = - if b a 1 a 2 b + if b a 1 a 2 b / G b (6.21)
With these substitutions one finds that the net rate of change
of phonon occupation in the state n1 n2 nb is determined by the
balance of annihilation versus creation.
{ }
.
+ + - - - - + + + - + -
n b = if nb n1 n2 - if b a 1 a 2 b + if b a 1 b 2 b a 2 b 1 (b b ) n2 n1 nb / G b
{
- - + + + + - -
}
- + - +
- if nb n1 n2 if b a 1 a 2 b - if b a 1 a 2 b a 2 a 1 (b b ) n2 n1 nb / G b
(6.22)
Upon evaluation of the matrix elements we obtain the rate
equation for phonon occupation.
nb = -( f 2/G b )n1 (n2 + 1)nb + ( f 2/G b )(n1 + 1)n2 (nb + 1) (6.23)
The first term on the right side of Eq. 6.23 describes cooling,
while the second describes heating. Possible detuning of the modes
has been ignored in this analysis. The phonon annihilation rate for
resonant Brillouin scattering in the interaction volume is therefore
the coefficient of the first term in Eq. 6.23.
γB = (f2/Γb)n1(n2 + 1)nb (6.24)
The energy loss rate takes into account the phonon frequency.
R = γBħΩ = (f2/Гb)n1(n2 + 1)nbħΩ (6.25)
Laser cooling is realized if conversion of an optical pump wave
to an anti-Stokes wave prevails over its conversion to a Stokes-
shifted wave [19]. Brillouin cooling of a mechanical mode was first
demonstrated in a microresonator by simultaneously matching
Raman Cooling 199
Ú
ε, or HI = 12 deE 2. This perspective facilitates the analysis of both
Brillouin and Raman scattering since it is straightforward to write δε
to reflect one interaction or the other. In the last section the dielectric
fluctuation in Brillouin scattering was written δε = –l(∂Q/∂x)to
describe electrostriction by the incident light. In this section δε
arises from dielectric fluctuations due to transverse charge motion
caused directly by the light field itself. We shall proceed as before
nevertheless, writing down the interaction Hamiltonian using
Eq. 6.9, since the deformation potential approach is equivalent to
that of Louisell.
For transverse optical modes, the net optical electric field can
act directly on charged ions at the difference frequency to create
or destroy optical phonons. In this case the potential is just the
optical energy density at the mode frequency. Hence the gradient of
the potential in Eq. 6.9 is —V = e —(E ◊ E * )e , and the single-mode
0 1 2
interaction Hamiltonian is
1/ 2
Ê e02 ˆ - + -
HI = Á ˜ xk xk + q (a 1 a 2 b + h.c .)
Ë 2WNM ¯
- + -
= f '(a 1 a 2 b + h.c .) (6.26)
where f ´ = –(2ħNMΩ)–1/2ε0Vq ξk ξk+q is the coupling strength in the
interaction volume V. By careful selection of the frequencies and
200 Methods for Laser Cooling of Solids
2
∆
ω1 ω2
δ
3
1
Figure 6.4 A three-level system undergoing a two-photon Raman process.
∆ =ω12 – ω is the one-photon detuning and δ =ω13 – (ω1 – ω2)
is the two-photon detuning.
2
∆
ω1 ω2
δ
3
1
Figure 6.5 Raman interaction of fields E1(ω1) and E2(ω2) in a time-
reversed (backward) sequence.
(1) Ê W12 / 2 ˆ
r12 =Á ( r11
(0)
- r22
(0)
) (6.31)
Ë D1 + iG12 ˜¯
The second-order coherence established by the forward Raman
process is obtained by solving the second-order transport equation
i r 13
( 2)
= (w1 - w3 )r13
( 2)
+ [V (1) , r(1) ]13 - i G 13 r13
(2)
(6.32)
The two-photon coherence oscillates at the frequency of
the (combined) driving fields. So if the slowly varying envelope
approximation is adopted, the form of ρ13 is
r13
( 2)
(t ) = r13
( 2)
exp[i(w1 - w 2 )t ] (6.33)
By substituting Eqs. 6.30, 6.31, and 6.33 into Eq. 6.32 and
grouping terms with the same time dependence, one finds the
solution for the coherence to be
Raman Cooling 203
ω3 ω4
Conducon
∆Laser
1 2 3
2
F7/2
∆F
2
F5/2
Figure 6.7 Energy levels and detunings involved in Raman laser cooling of
a three-level coolant atom in a solid. The example is based on
the electronic structure of Ce3+.
temperature. When Г’0 > Г0 the linewidth reduces to G 13 (T ) @< n > G '0.
Upon substitution into Eq. 6.35 with zero two-photon detuning (∆2
= 0), the rate of phonon annihilation per atom becomes constant.
2 2
gR = f M' i (6.39)
G '0
In this expression the modified matrix element M’ appears
because of the cancellation between <n> and two-photon linewidth
factors in the Orbach model outlined above. Its definition is
È 3 m (e ) ◊ e 2 2 2 m (e ) ◊ e 1 1 ˘ E(w )E *( -w )
f M' i = Í ˙ 1 2
(6.40)
ÍÎ [ D1 + iG 12 ] ˙˚ 4 2
The energy loss rate per atom
2 2
R = g R W = f M' i W (6.41)
G '0
is similarly expected to be temperature independent over the entire
range kBT > ħГ0. Unlike acoustic modes, optical phonons have
very little dispersion (Fig. 6.2). Consequently the frequency of the
dominant optical phonon does not vary significantly and R remains
constant over a wide range of temperature.
The 1D Raman interaction described above involves a phonon
near the center of the Brillouin zone propagating along a single optical
axis. However, zone center phonons cannot participate directly in
the collisional processes that establish thermal equilibrium within
the phonon distribution. Hence it is natural to inquire as to how
effectively a mode that is cooled irreversibly can equilibrate with the
reservoir of other phonons in the solid. Restoration of equilibrium
in solids is governed by Umklapp processes, phonon collisions
requiring a reciprocal lattice vector G to conserve momentum
[33]. Zone center phonons have low wavevectors, so when two
phonons interact with a third via an Umklapp process, there is no
way to satisfy the relation k1 + k2 = k3 + G with a wavevector G that
spans the entire Brillouin zone. To achieve uniform cooling of even
a single mode, phonons propagating along orthogonal axes should
be addressed, and this calls for the introduction of two more sets of
counterpropagating Raman beam pairs to cover all three orthogonal
space axes. Additionally, beam switching along each axis is needed
to interrogate both forward- and backward-traveling phonons of the
208 Methods for Laser Cooling of Solids
References
27. Reichel, J., Morice, O., Tino, G. M., Salomon, C. (1994). Subrecoil Raman
cooling of Cesium atoms, Europhys. Lett., 28, p. 477.
28. Reichel, J., Bardou, F., Ben Dahan, M., Peik, E., Rand, S. C., Salomon, C.,
Cohen-Tannoudji, C. (1995). Raman cooling of Cesium below 3 nK:
new approach inspired by Levy flight statistics, Phys. Rev. Lett., 75, pp.
4575–4578.
29. Rand, S. C. (2015). Lectures on Light, 2nd Ed. (Oxford University Press,
Oxford), Appendix G.
30. Kuhn, A., Perrin, H., Hansel, W., Salomon, C. (1997). Three dimensional
Raman cooling using velocity selective rapid adiabatic passage, in OSA
TOPS on Ultracold Atoms and BEC, 7, pp. 58–66.
31. Feofilov, S. P., Kulinkin, A. B., Gacoin, T., Mialon, G., Dantelle, G., Meltzer,
R. S., Dujardin, C. (2012). Mechanisms for Ce3+ excitation at energies
below the zero-phonon line in YAG crystals and nanocrystals, J. Lumin.,
132, pp. 3082–3088.
32. Abragam, A., Bleaney, B. (1970). Electron Paramagnetic Resonance of
Transition Ions (Dover, New York), pp. 560–562.
33. Kittel, C. (1971). Introduction to Solid State Physics, 4th Ed. (Wiley, New
York), p. 228.
Chapter 7
7.1 Introduction
Cooling of solid-state matter by laser radiation has definite
advantages over other types of cooling such as thermoelectric,
mechanical, and fluid. Optical coolers are compact, they have no
vibrations, and they use light as a coolant. Such coolers can be
effectively utilized in the case of extremely small solid-state objects
that are used in up-to-date optoelectronics. In the list of possible
applications of laser cooling in solids there are self-cooled lasers,
thermostabilized light-emitting diodes (LEDs), and space sensors.
Although different solid-state materials were proposed for laser
cooling in the last few decades, great success has been achieved to
date for glasses and crystals doped by the rare earth (RE) ions. There
are two reasons that ensure the success of laser cooling of such
materials. The first reason is a strong coupling between electrons
and vibrations in the RE ion. The ion electronic structure consists
of levels that are split into sublevels by a crystal field of the matrix
due to the Stark effect. The energy gaps between Stark-split sublevels
(SSSs) are comparable with phonon energies of the crystal matrix.
So, there are conditions to observe the anti-Stokes fluorescence
(ASF) in these solid systems. The second reason is the development
of technology that allows producing doped materials of high purity.
It means that in such RE ion systems there are low concentrations
of unwanted impurities. Reducing the impurity concentrations
decreases background absorption and unintended heating by a laser
radiation.
Laser cooling of a solid was originally attained in a Yb3+-doped
fluorozirconate glass that was cooled down by 0.3°С below room
temperature in 1995 [1]. Since that time, laser cooling was observed
in various glasses and crystals doped by the RE ions [2–7]. At present,
progress of the technology makes it possible to reach the range of
cryogenic temperatures for a Yb:YLF crystal [8]. This result is better
than the results obtained on cooling with the use of thermoelectric
Peltier elements.
The traditional model that describes the optical refrigeration of
doped solids is based on a four-level system of 4f multiplet states
of an RE ion [6]. In this model, the ASF is provided by (i) resonance
transitions between the upper ground and lower excited levels, (ii)
phonon absorption between the SSSs, and (iii) subsequent radiative
relaxation to the ground states. At the same time, in such a model,
the rate of the cooling cycle is very low because of the small cross
section of forbidden transitions between the levels of the same ion
configuration and because of the low population of the upper ground
level. For improving the cooling performance in the traditional
model, a nonresonant cavity for maximizing the absorbed laser
power is used [6].
In the last few decades, the coherent phenomena stimulated
by laser radiation, such as coherent population trapping (CPT) and
electromagnetically induced transparency (EIT), have been studied
in atomic physics most intensively [9]. At the same time, the CPT and
EIT effects, including the room-temperature case, were demonstrated
Introduction 213
for solid systems by several groups [10–12]. For the CPT effect to
be observed, it is very important to comply with certain relations
between the lifetimes of the electron levels involved in the cooling
process. For example, for a Λ system, in which there exist three levels
(the ground and first and upper excited levels), the presence of one
metastable level with a long lifetime (the first excited level) provides
optical interference phenomena. Such a situation is possible in RE-
doped crystals due to the existence of dipole-allowed 4f–5d and
forbidden 4f–4f optical transitions.
To improve the characteristics of the cooling process, such as
cooling efficiency, a minimum attainable temperature, and a cooling
rate, Raman mechanisms were recently proposed [13–15]. Raman
cooling used laser pulse pumping through dipole-allowed 5d ion
levels by the two-photon Raman (TPR) scattering proposed for
doped materials (see Section 6.4). The mechanism is based on 4f–5d
and 5d–4f optical transitions involving longitudinal optical phonons
of the crystal matrix. The acceleration of the cooling cycle is achieved
by combining the stimulated TPR process with third pulse pumping
at the allowed fluorescence transition. Nevertheless, in this model,
the cooling efficiency is 0.6%. This is because of the low transition
rate of the two-photon process that involves phonons. Moreover,
this cooling mechanism stops working after freezing the vibrations
of the optical modes. Another Raman mechanism of cooling is based
on methods of coherent pumping of doped crystals [14, 15]. The
coherent and complete population transfer between the ground
and the first excited levels of 4f multiplet through the 5d ion levels
was achieved by using the different Raman techniques, namely TPR
scattering, stimulated Raman adiabatic passage (STIRAP), and the
π-pulse method. It was shown in these works that the coherent
pumping increases the number of electrons that participate in the
cooling cycle, which leads to increasing of the cooling power. At
the same time, the results showed that the deformation potential
mechanism for description of the interaction between ion
electrons and vibrations of the crystal medium failed to explain the
establishment of the Boltzmann distribution of the electrons at the
SSSs.
In this chapter, we consider a cooling model with STIRAP for
optical pumping of an RE-ion-doped crystal, taking into account
vibronic interaction at the SSSs of ion levels. In the model, cooling
214 Deep Laser Cooling of Rare Earth–Doped Crystals by Stimulated Raman Adiabatic Passage
counterintuitive order when the first pulse initially couples the two
empty levels, |3Ò and |2Ò, and the second pulse couples levels |1Ò
and |3Ò. Such coherent pumping gives rise to the transfer of all the
population from level |1Ò to level |2Ò without populating level |3Ò [14].
The establishment of the Boltzmann distribution at levels |2Ò, |4Ò,
and |5Ò is the result of vibronic interaction. This interaction leads to
consumption of vibrational energy of the RE ions and consequently
to the cooling of the system. Finally, the population returns to the
ground level by radiative decay, and the cycle is completed.
Conduction band
|3〉 5d level
Δ1
Δ2
ω1 ω2
ω02 4f levels
|5〉
|4〉
|2〉
ω01
γ
|1〉
i ii iii
Figure 7.1 Scheme of the laser cooling cycle for the five-level energy
structure of an RE ion: (i) two-photon optical pumping, (ii)
vibronic interaction, and (iii) radiative decay. Solid arrows
show the optical transitions at the frequencies ω1 and ω2;
dotted arrows show the vibronic interaction with energies ω01
and ω02; downward dashed arrows refer to the nonradiative
decay; wavy arrows show the radiative decay γ; Δ1 and Δ2
are the detunings of optical transitions; the detunings of the
transitions under the vibronic interactions are not depicted
(see text).
+
23,l k a+ a exp(i D t ) + a+ a W
+ W 32 ,l2k 2 exp( -i D 2t )
2 2 3 2 2 2 3
+ Q + + +
24 ,01a4 a2exp(id 01t ) + a2 a4 Q 42,01exp( -id 01t )
+ Q + + +
45,02a5 a4exp(id 02t ) + a4 a5Q54 ,02exp( -id 02t ) , (7.2)
ij ,l k = mij E l k , Q
W kij
ij ,0n = b , (7.3)
n
m m m m
Here μij are the matrix elements of the dipole moment operator
+
for levels i and j; E lmk m ( E l k ) is the electric field operator that
m m
13,l k r + i r W
r 12 = i ( D1 - D 2 )r12 - i W + +
1 1 32 13 32,l2k 2 + i r14 Q 42,01 - g 12r12 ,
13,l k - i W
r 13 = i r11 W 13,l k r + i r W
1 1 1 1 33 12 23,l2k 2 + i D 1 r13 - g 13r13 ,
13,l k r + i r Q
r 14 = i (d 01 - D 2 + D1 )r14 - i W +
1 1 34 12 24 ,01 + i r15 Q54 ,02 - g 14 r14 ,
13,l k r + i r Q
r 15 = i (d 01 + d 02 - D 2 + D1 )r15 - i W
1 1 35 14 45,02 - g 15r15 ,
+
32
r 22 = i r23 W +
,l2k 2 - i W23,l2k 2 r32 + i r24 Q 42,01 - iQ 24 ,01 r42
+g 44
nr
r44 + g 33 r33 - g 22 r22 ,
23,l k - i W
r 23 = i D 2r23 + i r22 W 23,l k r - iQ
2 2 2 2 33 24 ,01 r43 + i r21 W13,l1k 1 - g 23r23 ,
+
r 24 = i d 01r24 + i r22 Q
24 ,01 - iQ 24 ,01 r44 - i W23,l2k 2 r34 + i r25 Q54 ,02 - g 24 r24 ,
nr
r 25 = i (d 01 + d 02 )r25 - iQ
24 ,01 r45 - i W23,l2k 2 r35 + i r24 Q 45,02 - g 25 r25 ,
nr
+
13,l k - i W
r 33 = i r31 W 31 +
,l1k 1 r13 + i r32 W23,l2k 2 - i W32,l2k 2 r23 - 4g 33r33 ,
1 1
Vibronic Model of Laser Cooling of Rare Earth Ions 219
+
31
r 34 = i (d 01 - D 2 )r34 - i W +
,l1k 1 r14 - i W32,l2k 2 r24
+
+i r32 Q
24 ,01 + i r35 Q54 ,02 - g 34 r34 ,
+
31
r 35 = i (d 01 + d 02 - D 2 )r35 - i W +
,l1k 1 r15 - i W32,l2k 2 r25 + i r34 Q 45,02 - g 35r35 ,
r 44 = i r42 Q + +
24,01 - iQ 42,01 r24 + i r45 Q54 ,02 - iQ 45,02r54
-(g 44 + g 44
nr
)r44 + g 55
nr
r55 + g 33 r33 ,
+ r + ir Q
r 45 = i d 02 r45 - iQ
42,01 25 44 45,02 - iQ 45,02 r55 - g 45 r45 ,
r55 = i r54 Q +
45,02 - iQ54 ,02 r45 - (g 55 + g 55 )r55 + g 33 r33. (7.4)
nr
1
J ( xk ) = d ( xk - nk ) (7.6)
2p nk
Here xk and xk are the eigenvalues and eigenstates of the
annihilation or the creation operator; nκ is the average photon
number in the mode. And the vibrational subsystem is described by
the density matrix operator
Ê ˆ
rv (0) = Z -1 exp Á -
ÁË Â b w
j
T
+
0 jbj bj ˜
˜¯
,
È Ê ˆ˘
Í ÁË j Â
Z = Tr Íexp Á - bT w0 j b+j b j ˜ ˙ , bT -1 = kBT .
˜¯ ˙
(7.7)
Î ˚
where kB is the Boltzmann constant and T is the temperature of
the system. In the case of a pseudolocal vibrational subsystem,
the averaging over the states is performed using the local density
of states (DOS) that is approximated by the δ-function for each
frequency ω01 and ω02 in our calculations. We assume the resonance
vibronic interaction that allows us to use the detunings δ01 = 0 and
δ02 = 0. Thus, taking into account the averaging procedure, we can
change the boson annihilation and creation operators by c-numbers
n exp(j ), where n is the average particle number in the mode and
φ is the phase of the eigenvalue of the annihilation or the creation
operator.
The time dependence of the diagonal elements of the reduced
density matrix for the first cooling cycle is shown in Fig. 7.2a. For
implementation of laser cooling we use two identical pulses that
have a Gaussian shape:
Ê (t - t0 )2 ˆ
E0 (t ) = E0 exp Á - ˜ . (7.8)
Ë s2 ¯
The width of each pulse is σ = 2.0 × 10–7 s and the electric field
amplitude of each pulse is E0 = 1.2 × 103 V/cm. The distance between
the pulses is 9.8 × 10–8 s (Fig. 7.2b). The first pulse couples the two
empty levels |3Ò and |2Ò, and the second pulse couples levels |1Ò and
|3Ò (see Fig. 7.1). The shapes and parameters of the pulses are chosen
to satisfy the adiabatic conditions: (i) a smooth pulse, (ii) a long
interaction time, and (iii) a large Rabi frequency [25]. Hence, the
Vibronic Model of Laser Cooling of Rare Earth Ions 221
1.0
0.8
0.6
0.4
0.2
0.0
–7 –7
0.0 2.0×10 4.0×10
–7 –7
0.0 2.0×10 4.0×10
Figure 7.2 (a) Time dependencies of the diagonal elements of the density
matrix: r11 (solid line), r22 (dashed line), r44 (dotted line), and
r55 (dash-dotted line). (b) Rabi frequencies of laser beams.
Note that the sequence of pulses is counterintuitive: first is
W32 (solid line) and second is W13 (dashed line). Results are
obtained at 300 K (cooling process C1).
222 Deep Laser Cooling of Rare Earth–Doped Crystals by Stimulated Raman Adiabatic Passage
5d 68741
ω1 ω2
2
F5/2
E7 10887
E6 10400
E 10209
2
F7/2
E4 686
E3 531
E2 110
E1 Yb 3+
0
Figure 7.3 Energy diagram of the Yb3+ ion and the laser radiation photon
energies used in calculations (in cm–1). Downward dashed
arrows denote the processes of radiative (nonradiative)
relaxation from the excited to ground sublevels. Only one set of
relaxation processes from level 7 is depicted (see text).
r 12 = -iQ +
13,l1k 1 r32 + i r13 Q32,l2k 2 + i r14 W 42,01 - g 12 r12 ,
nr
r 13 = i r11 Q +
13,l1k 1 - iQ13,l1k 1 r33 + i r12 Q23,l2k 2 - g 13 r13 ,
nr
r 14 = -iQ +
13,l1k 1 r34 + i r12W24 ,01 + i r15 Q54 ,02 - g 14 r14 ,
r 15 = -iQ
13,l1k 1 r35 + i r14 Q 45,02 - g 15r15 ,
r 22 = i r23 Q + +
23,l2k 2 - iQ23,l2k 2 r32 + i r24 W 42,01 - i W24 ,01 r42
+
r 23 = i r22 Q + ,
23,l2k 2 - iQ 23,l2k 2 r33 - i W24 ,01 r43 + i r21 Q13,l1k 1 - g 23 r23
nr
24 ,01 - i W
r 24 = i r22 W +
24 ,01r - iQ +
44 23,l2k 2 r34 + i r25 Q54 ,02 - g 24 r24 ,
r 25 = - i W +
24 ,01r - iQ
45 23,l2k 2 r35 + i r24 Q 45,02 - g 25r25 ,
r 33 = i r31 Q + +
13,l1k 1 - iQ31 ,l1k 1 r13 + i r32 Q23,l2k 2 - iQ32,l2k 2 r23
-g 33
nr
r33 + g 22
nr
r22 + g 44 r44 + g 55 r55,
+
r 34 = -iQ +
31 ,l1k 1 r14 - iQ32,l2k 2 r24 + i r32 W24 ,01 + i r35 Q54 ,02 - g 34 r34 ,
+
r 35 = -iQ
31 ,l1k 1 r15 - iQ32,l2k 2 r25 + i r34 Q 45,02 - g 35r35 ,
+
24 ,01 - i W
r 44 = i r42 W 42 +
,01 r24 + i r45 Q54 ,02 - iQ 45,02 r54 - 3g 44 r44 + g 55 r55 ,
nr
+
42
r 45 = -i W
,01 r25 + i r44 Q 45,02 - iQ 45,02 r55 - g 45 r45,
nr
Laser Cooling of the Yb3+:CaF2 System 225
r55 = i r54 Q +
45,02 - iQ54 ,02 r45 - 3g 55 r55 + g 55 r55 . (7.9)
nr
|5Ò
ω01
|4Ò
|2Ò
ω04
|3Ò
ω03
|1Ò
Figure 7.4 Scheme of vibronic interaction at the SSSs of the ground and
excited RE ion levels coupled by the electromagnetic field for
calculation of the number of electrons involved in the cooling
process. The five-level energy structure is used for simplicity
of calculations. The solid arrow shows the optical transitions
at the frequencies ω; dotted arrows show the vibronic
interaction with frequencies ω01, ω03, and ω04; and downward
dashed arrows refer to the nonradiative decay.
The changing of the level populations at the SSSs under the action
of the Gaussian laser pulse is depicted in Fig. 7.5. It is shown that
electrons from all SSSs of the ground level participate in the cooling
cycle at 300 K. Moreover, the aggregate population that is transferred
to the excited levels |4Ò and |5Ò after the pulse action is larger than
the population at the upper level |2Ò before the pulse action (Fig.
7.4). Thus, the existence of vibronic interaction at SSSs leads to an
increase of the number of electrons involved in the cooling cycle.
Since the electrons initially populated the ground levels |1Ò and |3Ò
consume energy to transfer to the upper level |2Ò, these electrons
participate in another cooling process that is additional to the
process described in Section 7.2. We denote this process as C2.
In the calculations, we use the following parameters: ħω03 = 421
cm–1; ħω04 = 155 cm–1; g14 = g15 = g24 = g25 = g34 = g35 = g/3; g55 = g44=
g; g22nr = g33nr = g55nr = gnr; g12nr = g13nr = g23nr = g45nr = 0.5 gnr; and T
226 Deep Laser Cooling of Rare Earth–Doped Crystals by Stimulated Raman Adiabatic Passage
1.00
0.95
0.90
0.85
0.80
–7 –7 –7
0.0 1.0×10 2.0×10 3.0×10
0.040
0.035
0.030
0.025
0.020
–7 –7 –7
0.0 1.0×10 2.0×10 3.0×10
Laser Cooling of the Yb3+:CaF2 System 227
0.08
0.06
0.04
0.02
0.00
–7 –7 –7
0.0 1.0×10 2.0×10 3.0×10
–7 –7 –7
0.0 1.0×10 2.0×10 3.0×10
S Ê Sˆ Ê Sˆ
e vib
C1
= r22
1
(t ) + r44
1
(t ) Á E6 - E5 + ˜ + r44
1
(t ) Á E7 - E5 + ˜ ,
4 Ë 4 ¯ Ë 4¯
(7.12)
e vib
C2
(
= r11
2
(0) - r11
2
(t ) ) (E 4 (
- E1 ) + r33
2
(0) - r33
2
(t ) ) (E 4 - E3 ) ,
(7.13)
( )
3
S ∫ 3E 4 - ÂE
m =1
m, N1 = nion Ne r44
2
(t ) + r55
2
(t ) . (7.14)
Ptot . (7.19)
hcool = -
Pabs
The absorbed power can be taken in the form
Pabs = N1ħ (ω1 – ω 2) λ. (7.20)
Finally, we must determine the power absorbed by impurity
ions. For this goal we can write the ratio of the power absorbed by
impurity ions to the power absorbed in the sample in Eq. 7.19 using
the Beer–Lambert law for both cases:
Pim 1 - exp( - b ) a
= , b = im , (7.21)
Pabs 1 - exp( -1) a rad
230 Deep Laser Cooling of Rare Earth–Doped Crystals by Stimulated Raman Adiabatic Passage
where αrad is the optical absorption coefficient for the RE ions and
αim is the absorption coefficient for the impurity ions. Equation 7.21
defines the absorption of laser radiation on the length of αrad–1.
It should be noted that the cooling process in the model
proposed in this study is basically more efficient than the process
involving forbidden optical transitions between the levels of the 4f
multiplet in the traditional model. To demonstrate this advantage of
the proposed model, we directly calculate the difference between
the cooling efficiencies in the above-mentioned two models using
the formulas available in Ref. [6]. Since the fundamental distinction
between the models is that pumping of the SSSs is implemented at
different (allowed and forbidden) optical transitions, the difference
between the cooling efficiencies is defined by the difference between
the absorption efficiencies only (all other parameters of the system
being the same). The absorption efficiency can be expressed as [6]
ηabs = (1 + β)–1. (7.22)
Then, we use the expression for the absorption coefficient to
find quantity β. The expression for the absorption coefficient can be
written as
αi = σiNi, (7.23)
where σi is the absorption cross-section for a certain wavelength
and Ni is the concentration of absorbing ions. If we assume that the
absorption cross sections for the RE and impurity ions are equal,
the quantity β is defined as the ratio of the concentrations of the RE
and impurity ions. Typical commercial fluoride starting materials
have transition metal impurities at the parts per million (parts
in 106) level [6]. And we assume that the crystal doped to the 5
mol% Yb content. So hence, the ratio of the absorption coefficients
is 2.0 × 10–5 in the case of allowed optical transitions. The optical
absorption coefficient controlled by dipole-allowed transitions in
the Yb3+ ion is 2 orders of magnitude larger than the absorption
coefficient controlled by dipole-forbidden transitions. Hence, by
determining the absorption efficiencies for the two models by Eqs.
7.22 and 7.23, we find that the cooling efficiency in the proposed
model with coherent pumping is higher by 0.005 (or by 14%) than
the cooling efficiency in the traditional model with pumping at
forbidden transitions (see Fig. 7.7).
Laser Cooling of the Yb3+:CaF2 System 231
to the SSSs of the excited level, is larger than the population of the
upper SSS of the ground level of the ion. The ratio of the difference
between these populations to the population of the upper SSS of the
ground level is defined as
r44
2
(t ) + r55
2
(t )
DN = - 1 . (7.27)
r22
2
(0)
7.4 Conclusions
A model of optical cooling of RE-doped crystals on coherent pulse
pumping of the SSSs of RE ions is considered in this chapter. It is
shown that the use of STIRAP through the dipole-allowed 5d
level of the RE ions improves the characteristics of the process of
laser cooling of solid systems in comparison to the corresponding
characteristics obtained on pumping at the forbidden 4f–4f
transitions. The absorption efficiency in the model proposed in the
study is 14% higher than the absorption efficiency in the traditional
model of direct 4f–4f pumping. In addition, in the proposed model,
the power and cooling rate of the system are increased at least by a
factor of 2 compared to these parameters in the traditional model.
This is due to the fact that the number of electrons involved in the
cooling cycle is larger in the case of coherent pumping than in the
case of direct 4f–4f pumping. As a result, it becomes possible to
implement deeper cooling for Yb3+-doped crystals with purity that
can be currently attained. This advantage of the proposed model
seems to be of particular importance. Our calculations show that
the minimum attainable temperature substantially depends on
the concentration of transition metal impurities in the case of the
traditional model. In this context, it is worth noting that the key
method of improving the cooling efficiency in the case of pumping
at forbidden optical transitions is to reduce the content of side
impurities in the crystal matrix, which presents a rather difficult
technological problem. Moreover, the proposed model is not
required to use the nonresonant cavity for maximizing the absorbed
laser power. Using coherent pumping, it becomes possible to reach
74.1 K in a time of 5 s. Since the optical cooling occurs at the electron
transitions between the SSSs, the limiting minimum temperature of
cooling in the above-developed model is the same as in other anti-
Stokes fluorescence (ASF) models [6].
This investigation shows importance of the nonadiabatic effects
for the problem of the laser cooling of solids doped by RE ions.
The vibronic interaction model used in our calculations specifies
the mechanism of interaction between the ion electrons and ion
vibrations in the crystal matrix. And the density matrix formalism
allows to describe coupling of the ion vibrations with the vibrations
of host material by introducing a mechanism of heat loss. In the
236 Deep Laser Cooling of Rare Earth–Doped Crystals by Stimulated Raman Adiabatic Passage
References
4. Patterson, W., Bigotta, S., Sheik-Bahae, M., Parisi, D., Tonelli, M., Epstein,
R. (2008). Anti-Stokes luminescence cooling of Tm3+ doped BaY2F8,
Opt. Express, 16(3), pp. 1704–1710.
5. Nemova, G., Kashyap, R. (2010). Laser cooling of Er3+-doped solids,
Opt. Commun., 283, pp. 3736–3739.
6. Seletskiy, D. V., Hehlen, M. P., Epstein, R. I., Sheik-Bahae, M. (2012).
Cryogenic optical refrigeration, Adv. Opt. Photonics, 4, pp. 78–107.
7. Fernandez, J., Garcia-Adeva, A. J., Balda, R. (2012). Anti-Stokes laser-
induced cooling in rare-earth doped low phonon materials, Opt. Mater.,
34, pp. 579–590.
8. Ghasemkhani, M., Albrecht, A. R., Melgaard, S. D., Seletskiy, D. V.,
Cedeberg, J. G., Sheik-Bahae, M. (2015). Intracavity-enhanced optical
refrigeration of Yb:YLF crystal to cryogenic temperatures, Proc. SPIE,
9380, p. 938003.
9. Fleischhauer, M., Imamoglu, A., Marangos, J. P. (2005).
Electromagnetically induced transparency: optics in coherent media,
Rev. Mod. Phys., 77, pp. 633–673.
10. Hemmer, P. R., Turukchin, A. V., Shahriar, M. S., Musser, J. A. (2001).
Raman-exited spin coherences in nitrogen-vacancy color centers in
diamond, Opt. Lett., 26(6), pp. 361–363.
11. Zhao, Y., Wu, C., Ham, B. S., Kim, M. K., Awad, E. (1997). Microwave
induced transparency in Ruby, Phys. Rev. Lett., 79(4), pp. 641–644.
12. Kolesov, R. (2005). Coherent population trapping in a crystalline solid
at room temperature, Phys. Rev., 72, p. 051801(R).
13. Rand, S. C. (2013). Raman laser cooling of solids, J. Lumin., 133, pp.
10–14.
14. Ivanov, A. V., Rozhdestvensky, Y. V. (2015). Laser cooling of doped
crystals by methods of coherent pumping, Proc. SPIE, 9380, pp.
93800S (1-10).
15. Ivanov, A., Rozhdestvensky, Y., Perlin, E. (2015). Coherent pumping for
fast laser cooling of doped crystals, J. Opt. Soc. Am. B, 32(5), pp. B47–
B54.
16. Perlin, Y. E., Kaminskii, A. A. (1985). Nonradiative transitions of the
trivalent lanthanides in insulating laser crystals, Phys. Status Solidi B,
132, pp. 11–40.
17. Lo, D., Makhov, V. N., Khaidukov, N. M., Krupa, J. C., Gesland, J. Y. (2004).
Upconverted VUV luminescence of Nd3+ and Er3+ doped into LiYF4
crystals under XeF-laser excitation, J. Lumin., 106, pp. 15–20.
238 Deep Laser Cooling of Rare Earth–Doped Crystals by Stimulated Raman Adiabatic Passage
8.1 Introduction
Optical cooling of solids, based on anti-Stokes fluorescence, has
tremendously advanced in the last years. Since its first experimental
demonstration in 1995 [5], the temperature drops achieved have
dramatically increased from a few degrees up to more than 150.
Operations in the cryogenic regime have been reached in ytterbium
(Yb)-doped fluoride single crystals [13, 14, 21]. To date, the lowest
temperature achieved is 93 K [12] and novel schemes based on
energy transfer processes prospect significant enhancements
0.2 .6 1.0 1.4 1.8 2.2 2.6 3.0 3.4 3.8 4.2 4.6 5.0 5.4 5.8 6.2 6.6 7.0 7.4
p(au)
Figure 8.1 Radial charge density of the 4f, 5s, 5p, and 6s shells of the Gd3+
free ion. Reprinted with permission from Freeman, A. J. and
Watson, R. E., Phys. Rev. (1893–1969), 127, 1962. Copyright
1962, by the American Physical Society.
THERMAL
SAMPLE VACUUM CHAMBER
CHAMBER
BaF2 window Thermocouple
SAMPLE
Mirror
Optical fibers
f = 10 cm f = 10 cm LD(1030 nm) (d = 120 mm)
glass window
BK7 windows
AR 1030 nm
LD (980 nm)
LD (1060 nm)
LD (940 nm)
SPECTROMETER
PC
optical fiber
E || c
E^c
(a)
YLF:10% Yb
YLF:7.5% Yb
YLF:5% Yb
(b)
Figure 8.3 (a) Polarized absorption coefficient of YLF:5%Yb acquired.
(b) Absorption coefficient of YLF:5%Yb, YLF:7.5%Yb, and
YLF:10%Yb for π-polarization.
254 Bulk Cooling Efficiency Measurements of Yb-Doped Fluoride Single Crystals
lf =
Ú I(l )ldl (8.5)
Ú I(l )dl
An excitation tuned to wavelengths longer than λf is required to
achieve anti-Stokes emission.
Polarized fluorescence spectra were acquired for the 5%, 7.5%,
and 10% Yb-doped YLF samples after laser excitation of the Yb
2F
5/2 manifold at 940 nm. Fluorescence spectra registered for the
YLF:5%Yb and YLF:10% samples are shown in Fig. 8.4.
Although the structure of emission spectra does not change
as the Yb concentration increases, the increase of reabsorption-
related effects for larger concentrations of active ions influences the
intensity distribution of the emission.
Due to the large overlap between absorption and fluorescence
bands, photons that are spontaneously emitted by excited ions
tend to be efficiently reabsorbed by other ions in the ground
state. Increasing the concentration of Yb ions, radiation-trapping
phenomena, due to reabsorption, tend to be more efficient, leading
to more depleted emissions at higher energies, as can be observed
by comparing spectra in Fig. 8.4.
Fluorescence spectra were used both to calculate the mean
emission wavelength parameter and the absorption coefficient via
the reciprocity method [17]. This technique is used to improve the
accuracy of the absorption coefficient measurement in the long-
wavelength tail of the spectrum, that is, the excitation region for
anti-Stokes emission. Due to the low value of the absorption in this
region, direct measurements normally exhibit large noise. Using
reciprocity theory, instead the resonant absorption is derived from
fluorescence spectra. The emission spectra exhibit a considerably
higher signal-to-noise ratio in this wavelength region and hence
improved accuracy for the absorption coefficient evaluation. Figure
8.5 reports, for comparison, the absorption coefficient of YLF:5%Yb
for the π-polarization calculated from direct measurements and
by using reciprocity. As can be seen from the graph, the reciprocity
theory well reproduces the behavior of the absorption coefficient
with significantly higher accuracy.
Experimental Results 255
E || c
E^c
(a)
E || c
E^c
(b)
Figure 8.4 Polarized fluorescence spectra of YLF:5%Yb (a) and
YLF:10%Yb (b), acquired, between 920 and 1100 nm with a
resolution 0.5 nm, after laser excitation of Yb ions at 940 nm
along the c axis.
256 Bulk Cooling Efficiency Measurements of Yb-Doped Fluoride Single Crystals
E || c
E^c
(a)
E || c
E^c
(b)
Figure 8.6 Polarized β-τ emission cross sections of YLF:5%Yb (a) and
YLF:10%Yb (b).
Table 8.1 EQE and background absorption values estimated for YLF
single crystals with varying Yb doping levels
Yb10%
Yb5%
5%
7.
Yb
Figure 8.8 Experimental data points and fit model curve of cooling
efficiency for Yb-doped YLF samples of variable doping levels.
Experimental Results 261
Yb
Yb
5%
%
10
5% Yb-Tm enriched
Figure 8.9 Experimental data points and fit model curve of Tm-enriched
YLF:5%Yb single crystal, along with cooling efficiency data
relative to single Yb doped at 5% and 10% YLF samples.
Experimental Results 263
(a)
(b)
Figure 8.10 Polarized absorption (a) and fluorescence (a) spectra of the
Yb transition acquired, in the same experimental conditions,
for the Tm-enriched YLF:5%Yb sample (labeled as A) and the
single doped YLF:5%Yb (labeled as B). Fluorescence spectra
were acquired by exciting Yb ions at 940 nm along the c axis.
Experimental Results 265
0.4
8
20X 0.3
6
0.2
4
2 0.1
0 0.0
460 480 500 520 540 560 620 640 660 680
wavelength [nm] wavelength [nm]
4.5
YLF:Yb5%-Tm0.0016% 0.40
4.0 YLF:Yb10%
YLF:Yb5%-Tm0.0016%
0.35 YLF:Yb10%
3.5
0.30
3.0
Intensity [a.u.]
0.25
2.5
2.0 0.20
1.5 0.15
1.0 0.10
0.5 0.05
0.0 0.00
720 740 760 780 800 820 840 860 1800 1900 2000 2100 2200
wavelength [nm] wavelength [nm]
the YLF samples doped only with Yb ions. The blue emission, between
450 and 500 nm, is dominant along with intense NIR fluorescence,
between 750 and 850 nm, and red fluorescence between 630 and
680 nm.
The atomic-like structure of the emissions in Fig. 8.14 suggested
trivalent rare earth impurities as the sources of such bands. Test
samples of YLF single crystals, doped with other rare earth ions, and
tabulated energies of Stark sublevels [8] were employed to identify
the elemental sources of the observed fluorescence.
The emission bands in the blue, red, and NIR region were
identified with the Tm transitions 1G4 Æ 3H6, 3H4 Æ 3H6, and 1G4 Æ
3F respectively. The green and 2 µm bands, which exhibited similar
4,
intensity in the co-doped and single Yb-doped YLF samples, instead,
were identified with combined emissions from Er and Ho ions.
The internal transitions within Tm ions that give rise to blue, red,
and NIR fluorescence hence appeared those involved in a virtuous
energy transfer process, favorable for the anti-Stokes cooling
mechanisms.
In the Yb–Tm system, energy transfer typically results in blue
upconversion via nonresonant three-photon processes [1]. These
mechanisms involve three sequential steps of nonradiative transfers
of excitation from excited Yb ions to Tm, with strong heat release
via multiphonon emission. However, when the Yb concentration is
dominant over Tm ions, a cooperative sensitization process between
Yb ions becomes more likely than nonresonant three-photon
upconversion, due to the strengthening of the Yb–Yb interaction.
In the cooperative sensitization process a pair of excited Yb ions
couple and transfer their energy to one Tm ion, which gets excited
to the 1G4 manifold [16, 18]. Such a mechanism is peculiar of the
Yb–Tm system because the energy of the acceptor state (Tm, 1G4)
is about twice the energy of the donor excited state (Yb, 2F5/2) and
nonradiative quenching of the acceptor state is prevented by the
absence of underlying states close enough in energy.
To identify which mechanism was dominant in the Yb–Tm co-
doped sample, the fluorescence intensity at 477 nm was measured
as a function of the pump power at 940 nm. The fit of experimental
data returned a power law with an exponent value of 2.3 ± 0.2,
which indicates that the excitation of the 1G4 manifold was mostly
Experimental Results 267
Acknowledgments
The authors would like to acknowledge I. Grassini for her competence
and care in preparing the samples. A. Volpi acknowledges support by
the European Space Agency under grant no. 4000108074/13/NL/
PA, “Cooling Effect on Fluoride Crystals.”
References
1. Auzel, F. (2004). Upconversion and anti-Stokes processes with f and d
ions in solids, Chem. Rev., 104, pp. 139–173.
2. De Loach, L. D., Payne, S. A., Chase, L. L., Smith, L. K., Kway, W. L., Krupke,
W. F. (1993). Evaluation of absorption and emission properties of Yb3+
doped crystals for laser applications, IEEE J. Quantum Electron., 29, pp.
1179–1191.
270 Bulk Cooling Efficiency Measurements of Yb-Doped Fluoride Single Crystals
3. Di Lieto, A., Sottile, A., Volpi, A., Zhang, Z., Seletskiy, D. V., Tonelli, M.
(2014). Influence of other rare earth ions on the cooling efficiency in
Yb:YLF crystals, Opt. Express, 22, pp. 28572–28583.
4. Dong, G. Z., Zhang, X. L., Li., L. (2013). Energy transfer–enhanced laser
cooling in Ho3+ and Tm3+ co-doped lithium yttrium fluoride, J. Opt.
Soc. Am. B, 30, pp. 939–944.
5. Epstein, R. I., Buchwald, M., Edwards, B., Gosnell, T., Mungan, C. (1995).
Observation of laser-induced fluorescent cooling of a solid, Nature,
377, pp. 500–502.
6. Hehlen, M. P., Epstein, R. I., Inoue, H. (2007). Model of laser cooling
in the Yb3+ -doped fluorozirconate glass ZBLAN, Phys. Rev. B, 75, p.
144302.
7. Hehlen, M. P., Boncher, W. L., Melgaard, S. D., Blair, M. W., Jackson, R. A.,
Littleford, T. E., Love, S. P. (2014). Preparation of high-purity LiF, YF3
and YbF3 for laser refrigeration, Proc. SPIE, 9000, p. 900004.
8. Kaminski, A. A. (1996). Crystalline Lasers: Physical Processes and
Operating Schemes (CRC Press).
9. Kastler, A. (1950). Some suggestions concerning the production and
detection by optical means of inequalities in the populations of levels
of spatial quantization in atoms. Application to the Stern and Gerlach
and magnetic resonance experiments, J. Phys. Radium, 11, pp. 255–
265.
10. Kuhn, H., Fredrich-Thornton, S. T., Krankel, C., Peters, R., Petermann, K.
(2007). Model for calculation of radiation trapping and description of
the pin-hole method, Opt. Lett., 32, pp. 1908–1910.
11. Melgaard, S. D. (2013). Cryogenic Optical Refrigeration: Laser Cooling
of Solids Below 123K, PhD dissertation, University of New Mexico,
Albuquerque, NM.
12. Melgaard, S. D., Albrecht, A., Hehlen, M. P., Seletskiy, D. V., Sheik-Bahae,
M. (2014). Optical refrigeration cools below 100K CLEO 2014, OSA
Tech. Dig., paper FTh4D.4.
13. Melgaard, S. D., Seletskiy, D. V., Di Lieto, A., Tonelli, M., Sheik-Bahae,
M. (2013). Optical refrigeration to 119K, below National Institute of
Standards and Technology cryogenic temperature, Opt. Lett., 38, pp.
1588–1590.
14. Melgaard, S. D., Seletskiy, D. V., Polyak, V., Asmerom, Y., Sheik-Bahae, M.
(2014). Identification of parasitic losses in Yb:YLF and prospects for
optical refrigeration down to 80K, Opt. Express, 22, pp.7756–7764.
References 271
15. Nemova, G., Kashyap, R. (2011). Alternative technique for laser cooling
with super irradiance, Phys. Rev. A, 83, p. 013404.
16. Ovsyakin, V. V., Feofilov, P. P. (1966). Cooperative sensitization of
luminescence in crystal activated with rare earth ions, Sov. Phys. JETP
Lett., 11, pp. 317–318.
17. Payne, S. A., Chase, L. L., Smith, L. K., Kway, W. L., Krupke W. F. (1992).
Infrared cross-section measurements for crystal doped with Er3+,
Tm3+ and Ho3+, IEEE J. Quantum Electron., 28, pp. 2619–2630.
18. Pei., X., Hou, Y., Zhao, S., Xu, Z., Teng, F. (2005). Frequency upconversion
of Tm3+ and Yb3+ codoped LiYF4 synthesized by hydrothermal
method, Mater. Chem. Phys., 90, pp. 270–274.
19. Ruan, X. L., Kaviany, M. (2006). Enhanced laser cooling of rare earth ion
doped nanocrystalline powders, Phys. Rev. B, 73, p. 155422.
20. Seletskiy, D. V., Hehlen, M. P., Epstein, R., Sheik-Bahae, M. (2012).
Cryogenic optical refrigeration, Adv. Opt. Photonics, 4, pp. 78–107.
21. Seletskiy, D. V., Melgaard, S. D., Bigotta, S., Di Lieto, A., Tonelli, M., Sheik
Bahae, M. (2010). Laser cooling of solids to cryogenic temperatures,
Nat. Photonics, 4, pp. 161–164.
22. Sheik-Bahae, M., Epstein, R. I. (2007). Optical refrigeration, Nat.
Photonics, 1, pp. 693–699.
23. Tiepolo, B., Bottazzi, P., Palenzona, M., Vannucci, R. (2003). A laser probe
coupled with ICP: double focusing sector; field mass spectrometer
in situ analysis of geological samples and U-Pb dating of zircon, Can.
Mineral., 41, pp. 259–272.
24. Volpi, A., Di Lieto, A., Tonelli, M. (2015). Novel approach for solid state
cryocoolers, Opt. Express, 23, pp. 8216–8226.
Chapter 9
B. Rami Reddy
Alabama A&M University, Department of Physics, Normal,
AL 35762, USA
rami.bommareddi@aamu.edu
and the paths travelled by both the beams are equal; the beams
form a rectangle or a parallelogram. Thus a collimated input beam
generates two wavefronts. If the wavefronts make a small angle with
each other a nonlocalized interference pattern is seen, consisting of
equispaced straight fringes parallel to the line of intersection of the
wavefronts. If there is an angle between the beams after exiting the
second beam splitter then localized fringes are formed. On the other
hand if there is no angle between the beams then nonlocalized fringes
are formed. The Mach–Zehnder interferometer is a much more
versatile instrument than the Michelson interferometer because
the optical beams are widely separated [20–23]. The beams travel
the path only once and the fringes can be localized at any desired
point. Because of this, it has been extensively used in studies of gas
flow, combustion, plasma density, and diffusion, where changes in
refractive index occur, which can be related to changes in pressure,
temperature, or the relative concentrations of different components
of a mixture [13, 23]. However, adjustment of the interferometer
is tedious to get fringes of good visibility, since a displacement of a
mirror results in a shift of the plane of localization as well as a change
in the optical path difference (OPD). The need for this procedure can
be eliminated by using pentaprisms in place of mirrors.
N1 g1 -DE10 /kT
= e (9.3)
N0 g0
3
2
l21
1
Laser
0
Figure 9.1 Energy levels and relaxation phenomena in a rare earth ion–
doped solid. Radiative and nonradiative relaxation phenomena
are represented by downward straight-line arrows and wavy
arrows. Emission from level 3 (dotted line) occurs if its
population is sustained by thermalization.
10 200 cm-1
2F
5/2
9 800
700
2F
7/2
0
Figure 9.4 Partial energy level diagram of YAG:Yb3+. Upward and
downward arrows represent laser excitation and fluorescence.
Wavy lines represent nonradiative relaxation.
detector
screen
heat cool
computer
514.5 nm
BS2
M1 915 nm
YAG: Yb3+
BS1 M2
2
514.5 nm
Figure 9.5 A 915 nm laser beam heats the YAG:Yb3+ sample. A 514.5 nm
beam was used to setup a Mach–Zehnder interferometer to
monitor temperature changes [36].
284 Interferometric Measurement of Laser-Induced Temperature Changes
Time (s)
dt. The path lengths of beams 1 and 2 are η1(L2 + L1 – L3) + η2L3 and
η1(L1 + L2), where η1 and η2 are refractive indices of air (η1 = 1.0)
and YAG (η2 = 1.823), respectively; L1 and L2 are the dimensions of
the rectangle; and L3 is the sample length. The OPD between the two
beams 1 and 2 is given by OPD = (η2 – η1)L3. Here, η2 and L3 are
dependent on sample temperature.
The refractive index and sample length are temperature
dependent, which are described by
∂h
h3 = h0+ DT (9.8)
∂T
∂L
L3 = L0+ DT (9.9)
∂T
where η0 and L0 are the room-temperature values of refractive index
and sample length, respectively. When the Ti:sapphire laser beam
passes through the sample its temperature changes from T1 to T2.
However, the change in OPD is given by
Δ(OPD) = Δ(η2L3 – η1L3) (9.10)
Since η1 = 1.0 for air,
Δ(OPD) = Δ(η2L3) – Δ(η1L3) = Δη2L3 + η2ΔL3 – ΔL3 (9.11)
∂h ∂L ∂L
D(OPD) = L3 DT + h3 DT - DT (9.12)
∂T ∂T ∂T
Hereafter L3 = L0 = L (sample length) and η2 = η0 = η (sample
refractive index).
∂h ∂L ∂L
D(OPD) = L DT + h DT - DT (9.13)
∂T ∂T ∂T
Ê 1 ∂h 1 ∂L 1 ∂L ˆ
D(OPD) = Lh Á DT + DT - DT ˜ (9.14)
Ë h ∂T L ∂T Lh ∂T ¯
Equation 9.14 can be written as
Ê 1 ∂h aˆ
D(OPD) = Lh Á + a - ˜ DT (9.15)
Ë h ∂T h¯
where α is the linear thermal expansion coefficient of the material,
given by
1 ∂L
a= (9.16)
L ∂T
286 Interferometric Measurement of Laser-Induced Temperature Changes
where
1 ∂h
g = +a (9.18)
h ∂T
∂T l N 1
=- (9.22)
∂L hÊ a ˆ L2
ÁË g - h ˜¯
∂T -gl N 1
= (9.23)
∂h L Ê aˆ h
2 2
ÁË g - h ˜¯
Mach–Zehnder Interferometer 287
∂T l N 1
= (9.24)
∂a L Ê aˆ h
2 2
ÁË g - h ˜¯
∂T l N h
= (9.25)
∂g L Ê aˆ h
2 2
ÁË g - h ˜¯
Time (s)
Figure 9.8 Temporal variation of the detector output during cooling (915
nm laser was turned off at t = 0 s).
10.5
Laser off (cooling)
Intensity (a.u.)
8.5
6.5
Laser on (heating)
4.5
0 500 1000 1500 2000
time (s)
Figure 9.9 Temporal evolution of the detector output during the time of
laser heating (0 to 1000 s) and laser off/cooling (1000 to 2000
s).
where
IRF = I0 cos[2πΔνt + ΔΘ(z)] (9.28)
where ΔΘ(z) = Θ1(z1) – Θ2(z2). Θ1(z1) is the phase of the reference
beam and Θ2(z2) is the phase of the measurement beam. As the
material is cooled (or heated) the optical path length travelled by
the measurement beam changes and, hence, its phase, Θ2(z2), also
changes. Consequently the phase difference between the two waves,
ΔΘ(z) changes continuously so long as the material temperature
changes. ΔΘ(z) is related to the path length change Δz as
ΔΘ(z) = (2π/λ)Δz (9.29)
where λ is the wavelength corresponding to the center frequency of
the Zeeman laser.
6 ×1014 Hz
ν1 dc current
ν1
ν2 dc current + rf frequency
rf frequency = ν2 - ν1 = 250 kHz
PBS
L1 M1
V1
Zeeman Reference
Computer
laser (rf)
L2
V2
h2 detector
DM
M2
rf
L3 POL Phase meter
Ti:Sapphire laser
Ê 1 ∂h 1 ∂L 1 ∂L ˆ
D (OPD) = hL Á + - DT (9.37)
Ë h ∂T L ∂T hL ∂T ˜¯
where α and γ have their usual meaning. The change in the OPD is
simplified as
Ê aˆ
D(OPD) = hL Á g - ˜ DT (9.38)
Ë h¯
The OPD is related to the phase difference between the two
beams as
Optical Heterodyne Technique 293
2p 2p Ê aˆ
DQ = D(OPD) = hL Á g - ˜ DT (9.39)
l l Ë h¯
By rearranging the above equation, the temperature difference is
estimated as [44]
lDQ
DT = (9.40)
Ê aˆ
2phL Á g - ˜
Ë h¯
(9.41)
where the subscript “u” refers to the uncertainty in the parameter
and the partial derivatives are evaluated from Eq. 9.40. The uncer-
tainty in wavelength, Δλu is zero because the probe laser wavelength
does not change. Therefore, the accuracy in ΔT estimation depends
on the uncertainties in the estimates of η, α, g, and ΔΘ measurement.
∂T l Dq d
= (9.42)
dq 2p Ê aˆ
ÁË g - Lh
h ˜¯
∂T l qd DL
=- (9.43)
dL 2p Ê a ˆ L2
ÁË g - h
h ˜¯
294 Interferometric Measurement of Laser-Induced Temperature Changes
∂T l qd 1
=- (9.44)
dh 2p Ê a ˆ h2
ÁË g - h ˜¯
∂T l qd 1
=- 2 (9.45)
da 2p Ê a ˆ Lh
2
ÁË g - h ˜¯
∂T l qd 1
= 2 Lh (9.46)
dg 2p Ê aˆ
ÁË g - h ˜¯
sample collinearly with the He–Ne laser. The color filter blocks
the heating beam from reaching the screen. Both the length and
the refractive index of a material change whenever it is heated. A
measurement of optical path length variation of the beam that
passes through the material reveals the sample temperature change.
So, prior knowledge of the temperature dependence of the material’s
refractive index and thermal expansion coefficient is needed [38].
The results obtained for Pr3+-doped YAG are summarized here [2].
He–Ne
Ar+
SA M3
BS2 BS1
M2
CF
Lens
SA sample Heat
BS beam splitter
M mirror Cool
D detector
CF color filter M1
Computer
3P
2
3
P0
1D
2
1
G4
617 nm
488 nm
3H
6
3H
4
Figure 9.14 Temporal variation of the detector output obtained when 115
mW of an Ar+ laser beam was launched through the sample.
The sample was located at the focus. The Ar+ laser was on from
t = 0 to 600 s [2].
Figure 9.15 Temporal variation of the detector output during heating (Ar+
laser was on from t = 0 to 135 s) and cooling (Ar+ laser was off
from t = 135 to 300 s) periods [2].
Michelson Interferometer 299
fast to reach the set value. However, there was some time lag (couple
of minutes) before the thermocouple senses higher temperatures,
because of the delay involved in heating the water. When the
temperature started to increase the computer was manually
triggered for data acquisition. This process has generated some
error in timing the events.
Figure 9.17 Temporal variation of the detector output measured when the
sample holder was (a) heated in a water bath from 22.5°C (t =
0 s) to 36.5°C (t = 853 s) and (b) naturally cooled in ambient air
from 35°C (t = 0 s) to 27.6°C (t = 2400 s).
References
4. Soto, M. A., Nannipierri, T., Signorini, A., Lazzeri, A., Baronti, F., Roncella,
R., Bolognini, G., Pasquale, F. D. (2011). Raman-based distributed
temperature sensor with 1 m spatial resolution over 26 km SMF using
low repetition-rate cyclic pulse coding, Opt. Lett., 36, pp. 2557–2559.
5. Lapp, M., Hartley, D. L. (1976). Raman scattering studies of combustion,
Combust. Sci. Technol., 13, pp. 199–210.
6. Wade, S. A., Collins, S. F., Baxter, D. W. (2003). Fluorescence intensity
ratio technique for optical fiber point temperature sensing, J. Appl.
Phys., 94, pp. 4743–4756.
7. Chubb, D. L., Wolford, D. S. (2000). Rare-earth optical temperature
sensor, Rev. Sci. Instrum., 71, pp. 2233–2237.
8. Zhan, Y., Wu, H., Yang, Q., Xiang, S., He, H. (2008). Fiber Bragg grating
sensors for high temperature measurement, Opt. Lasers Eng., 46, pp.
349–354.
9. de Lima Filho, E. S., Nemova, G., Loranger, S., Kashyap, R. (2014).
Direct measurement of laser cooling of Yb:YAG crystal at atmospheric
pressure using a fiber Bragg grating, Proc. SPIE, 9000, p. 90000I.
10. Hariharan, P. (2003). Optical Inteferometry (Academic Press, San
Diego, USA).
11. Yoshizawa, T. (2009). Handbook of Optical Metrology: Principles and
Applications (CRC Press, Boca Raton, USA).
12. Wilkie, D., Fisher, S. A. (1963). Measurement of temperature by Mach-
Zehnder interferometry, Proc. Inst. Mech. Eng., 178, pp. 461–470.
13. Sandhu, S. S., Weinberg, F. J. (1972). A laser interferometer for
combustion, aerodynamics and heat transfer studies, J. Phys. E: Sci.
Instrum., 5, pp. 1018– 1020.
14. Steel, W. H. (1983). Interferometry (Cambridge University Press, New
York, USA).
15. Malacara, D. (1992). Optical Shop Testing (John Wiley & sons, New
York, USA).
16. Toal, V. (2012). Introduction to Holography (CRC Press, New York,
USA).
17. Hariharan, P. (1969). Modified Mach-Zehnder interferometer, Appl.
Opt., 8, pp. 1925–1926.
18. Kowalski, F. V., Teets, R. E., Demtröder, W., Schawlow, A. L. (1978). An
improved wavemeter for cw lasers, J. Opt. Soc. Am., 68, pp. 1611–1613.
19. Lee, L. S., Schawlow, A. L. (1981). Multiple-wedge wavemeter for
pulsed lasers, Opt. Lett., 6, pp. 610–612.
References 303
20. Luo, H., Sun, Q., Xu, Z., Liu, D., Zhang, L. (2014). Simultaneous
measurement of refractive index and temperature using multimode
microfiber-based dual Mach–Zehnder interferometer, Opt. Lett., 39,
pp. 4049–4053.
21. Lu, P., Chen, Q. (2010). Asymmetrical fiber Mach–Zehnder
interferometer for simultaneous measurement of axial strain and
temperature, IEEE Photonics J., 2, pp. 942–952.
22. Dullo, F. T., Sokolov, V., Chauvet, C., Lindecrantz, S., Solbø, S. A., Hellesø,
O. G. (2015). Application of Mach–Zehnder interferometer based
on long period grating structure for salinity measurement in water
environment, Proc. SPIE, 8988 89881T-1.
23. Ashrafi, Z. N., Ashjaee, M., Askar, M. H. (2015). Two-dimensional
temperature field measurement of a premixed methane/air flame
using Mach–Zehnder interferometry, Opt. Commun., 345, pp. 55–63.
24. Koechner, W., Bass, M. (2003). Solid-State Lasers (Springer, New York,
USA).
25. Moerner, W. E. (1988). Persistent Spectral Hole Burning: Science and
Applications, Topics in Current Physics, Vol. 44 (Springer, New York,
USA).
26. Epstein, R. I., Sheik-Bahae, M. (2009). Optical Refrigeration: Science and
Applications of Laser Cooling Of Solids (Wiley VCH Verlag, Weinheim).
27. Dieke, G. H. (1968). Spectra and Energy Levels of Rare-Earth Ions (John
Wiley & Sons, USA).
28. Weber, M. J. (1967). Probabilities for radiative and nonradiative decay
of Er3+ in LaF3, Phys. Rev., 157, pp. 262– 272.
29. Riseberg, L. A., Moos, H. W. (1967). Multiphonon orbit-lattice relaxation
in LaBr3, LaCl3 and LaF3, Phys. Rev. Lett., 19, pp. 1423–1426.
30. Kamma, I., Kommidi, P., Reddy, B. R. (2008). Design of a high
temperature sensing system using luminescent lifetime measurement,
Rev. Sci. Instrum., 79, p. 096104.
31. Reddy, B. R., Kamma, I., Kommidi, P. (2013). Optical sensing techniques
for temperature measurement, Appl. Opt., 52, pp. B33–B39.
32. Messias, D. N., Vermelho, M. V. D., Gouveia-Neto, A. S., Aitchison, J.
S. (2002). All optical integrated upconversion fluorescence based
point temperature sensing system using Er3+ doped silica-on-silicon
waveguides, Rev. Sci. Instrum., 73, pp. 476–479.
33. Paez, G., Strojnik, M. (2003). Erbium-doped optical fiber fluorescence
temperature sensor with enhanced sensitivity, a high signal-to-noise
304 Interferometric Measurement of Laser-Induced Temperature Changes
ratio and a power ratio in the 520-530- and 550-560 nm bands, Appl.
Opt., 42, pp. 3251–3258.
34. Paez, G., Strojnik, M. (2003). Experimental results of ratio-based
erbium-doped-silica temperature sensor, Opt. Eng., 42, pp. 1805–
1811.
35. Patterson, W., Soto, E., Fleharty, M., Sheik-Bahae, M. (2007).
Differential luminescence thermometry in laser cooling of solids, Proc.
SPIE., 6461, p. 64610B.
36. Farley, C. W., Reddy, B. R. (2011). Mach-Zehnder interferometric
measurement of laser heating/cooling in Yb3+: YAG, Proc. SPIE, 7951,
p. 795101.
37. YAG data sheet, VLOC, FL (www.vloc.com).
38. Fan, T. Y., Daneu, J. L. (1998). Thermal coefficients of the optical path
length and refractive index in YAG, Appl. Opt., 37, pp. 1635–1637.
39. Wynne, R., Daneu, J. L., Fan, T. Y. (1999). Thermal coefficients of the
expansion and refractive index in YAG, Appl. Opt., 38, pp. 3282–3284.
40. Sommergen, G. E. (1981). Optical heterodyne profilometry, Appl. Opt.,
pp. 610–618.
41. Huang, C. C. (1984). Optical heterodyne profilometer, Opt. Eng., 23, pp.
365–370.
42. George, M. C., Reddy, B. R., Jagannath, H., Perera, G. M., Venkateswarlu,
P. (1991). Measurement of surface roughness using a noncontact
profilometer, in Laser Spectroscopy and Nonlinear Optics of Solids,
Radhakrishna, S., Tan, B. C., eds. (Springer Verlag/Narosa, New Delhi,
India), pp. 147–152.
43. Kim, Y. K., Reddy, B. R., George, T. G., Lal, R. B. (1998). Optical
heterodyne interferometry technique for solution crystal growth rate
measurement, Opt. Eng., 37, pp. 616–621.
44. Reddy, B. R. (2014). Applications of optical interferometer techniques
for precision measurements of changes in temperature, growth and
refractive index of materials, Technologies, 2, pp. 54–75.
45. Taylor, J. R. (1997). An Introduction to Error Analysis: The Study of
Uncertainties in Physical Measurements, 2nd Ed. (University Science
Books, Sausalito, CA).
46. Kamma, I., Kommidi, P., Reddy, B. R. (2009). High temperature
measurement using luminescence of Pr3+ doped YAG and Ho3+ doped
CaF2, Phys. Status Solidi C, 6, pp. S187–S190.
Chapter 10
Mohammed Saad
Thorlabs, Inc., 56 Sparta Avenue, Newton, NJ 07860, USA
msaad@thorlabs.com
10.1 Introduction
During the last three decades, optical fibers have experienced
an extraordinary and rapid development. They are being used, as
passive and active mediums, in various industrial applications,
in fields as diverse as telecommunication, medicine, aerospace,
defense, spectroscopy, sensing, laser power delivery, fiber lasers, etc.
Optical fibers are now critical parts in many high-tech devices.
Of course fiber parameters have to be tailored to each application
needs. Unfortunately, there is no single fiber material that can fulfill
all types of application needs. Engineers and researchers have to
make some compromises when choosing the right fiber material
for the right application. There are some important criteria that
one has to consider when choosing an optical fiber to be used in a
specific application. The first criterion is technical. The glass should
have the required properties, such as transmission window, glass
This process is only limited to binary thin films and can’t be used to
prepare bulk glasses and preforms.
As far as fluoride glasses are concerned, the preparation process
commonly used is melting and casting. In this case ultrahigh-purity
starting materials are required. And raw materials’ purity is a big
concern.
In fact, many chemical molecules and ions, such as transition
metals [22], rare earth elements [23], water [24], CO2 [25], and
some organic molecules, have their fundamental absorption in the
transmission window of fluoride glasses. And only a very small
amount of these impurities in the glass will have a huge absorption
peak in the fiber spectrum. Thus, from the beginning the purity of
raw materials was a critical issue. At that time no ultrahigh-purity
raw materials were available, especially those entering stable
glass compositions. As an example, Table 10.2 shows attenuation
induced by some transition metals and rare earth elements [22]. As
an example, –OH and transition metal ion concentration has to be
kept in the ppb level. There are two challenges here: The first one
is to be able to produce the material with a low impurity level, and
the second is to be able to measure this impurity level with good
accuracy. Another challenge is that many raw material suppliers are
not really interested in making some chemicals with such ultrahigh-
purity materials since the market is too small. So sometimes you
have to produce or purify your own raw materials.
Elements 2 mm 2.5 mm 3 mm
Transition metals (dB/km/ppm) (dB/km/ppm) (dB/km/ppm)
Fe 90 28 2
Co 130 31 4
Ni 90 6 0.5
Cu 3 0.14 0.01
Rare earth
Ce – – 7.7
Pr 27 1.8 0.4
Tb 25 – 14.3
312 Fluoride Glasses and Fibers
the glass formation region and the stable composition region have
been defined using the stability criterion, one can use the CCR to
determine the most stable composition of the system among all stable
compositions. Table 10.3 shows the CCR values for the two standard
fluorozirconate glasses ZBLAN and ZBLA. The ZBLAN composition
has the lowest CCR among all fluoride glass compositions reported
so far [36]. In fact, after melting, the ZBALN composition glass can be
cooled slowly in the crucible and no crystallization occurs.
Glass Rc [°C/s]
ZBLAN 0.13
ZBLA 0.28
70
60
50
40
30
20
10
0
0 1 2 3 4 5 6 7 8 9 10
Wavelength (mm)
Figure 10.4 ZBLAN convex and concave lenses made by the single-point
diamond turning technique.
m0 l 2 m1 l 2 m2 l 2
n( l ) = + + +A (10.3)
l 2 - m32 l 2 - m42 l 2 - m52
ZrF4 Glass
Wavelength l (nm) Refractive index nl
405 1.506
486.1 1.504
587.6 1.4995
632.8 1.49807
656.3 1.4974
1060 1.49198
1550 1.48883
3500 1.47447
4500 1.46312
InF3 Glass
Wavelength l Refractive index nl
4500 1.45886
2000 1.47384
1064 1.47828
587.6 1.48545
485 1.49005
404.7 1.49645
320 Fluoride Glasses and Fibers
Coefficient Cladding
m0 0.705674
m1 0.515736
m2 2.204519
m3 0.087503
m4 0.087505
m5 23.80739
A 1
Coefficient Cladding
m0 0.68462594
m1 0.4952746
m2 1.4841315
m3 0.0680833
m4 0.11054856
m5 24.4391868
A 1
Figure 10.8 A 125 um InF3 fiber’s diameter variation during the drawing
process.
324 Fluoride Glasses and Fibers
50.00
Unreliability, F(t)
10.00
5.00
MSaad
Irphotonics
1.00 11/9/2011 09:17
10.00 100.00 1000.00
Stress at Breakage (kpsi)
Figure 10.13 Weibull plot for InF3 multimode fiber tensile strength.
90.00 W2 RRX-SRMMED
F=30 / S= 0
50.00
Unreliability, F(t)
10.00
5.00
Patrick Orsini
IR Photonics
2010-09-07 13:09
1.00
10.00 100.00 1000.00
Stress at breakage (kpsi)
b=5.8657, h=236.0039, r=0.9743
References
1. Yu, F., Wadsworth, W. J., Knight, J. C. (2012). Low loss silica hollow core
fibers for 3–4 μm spectral region, Opt. Express, 20(10), pp. 11153–
11158.
2. Poulain, M. (1995). New trends in halide glass compositions, J. Non-
Cryst. Solids, 184, pp. 103–108.
3. Conseil, C., Bastien, J.-C., Boussard-Plédel, C., Zhang, X.-H., Lucas, P., Dai,
S., Lucas, J., Bureau, B. (2012). Te-based chalcohalide glasses for far-
infrared optical fiber, Opt. Mater. Express, 2(11), pp. 1470–1477.
4. Wang, J. S., Vogel, M. E., Snitzer, E. (1994). Tellurite glass: a new
candidate for fiber devices, Opt. Mater., 3, pp. 177–203.
5. Guorong, C., Wen, L., Jijian, C. (2000). Improved properties for precision
molding phosphate glasses, Am. Ceram. Bull., pp. 44–46.
6. Feigelson, R. S., Kway, W. L., Route, R. K. (1984). Single crystal fiber by
the laser-heated pedestal growth method, Proc. SPIE, 0484 (Infrared
Optical Materials and Fibers III), p. 133.
7. Saad, M. (2009). Fluoride glass fiber: state of the art, Proc. SPIE, 7316,
p. 73160N.
8. Van Uitert, L. G., Wemple, S. H. (1978). ZnCl2 glass: a potential ultralow-
loss optical fiber materials, Appl. Phys. Lett., 33, pp. 57–58.
9. Poulain, M., Poulain, M., Lucas, J., Brun, P. (1975). Verres flores au
tetrafluorure de zirconium proprietes optiques d’un verre dope au
Nd3+, Mater. Res. Bull., 10, pp. 243–246.
10. Tregoat, D., Fonteneau, G., Lucas, J. (1985). Verres fluores a base de
metauxlours. Traitement des impuretes oxydes et hydroxides sous
atmosphere controlee, Mater. Res. Bull., 20, pp. 179–185.
11. Robinson, M. (1985). Preparation and purification of fluoride glass
starting materials, Mater. Sci. Forum, 5–6, pp. 19–33.
12. Tran, D. C., Sigel, G. H., Bendow, B. (1984). Heavy metal fluoride glasses
and fiber, a review, J. Lightwave Technol., LT2(5), pp. 566–586.
13. Nakai, T., Mimura, Y., Hokiwa, H., Shinbori, O. (1986). Dehydration of
fluoride by NF3 processing, J. Lightwave Technol., LT4, pp. 87–89.
14. Adan, J. L., Ricordel, C., Lucas, J. (1997). New composition of low
phonon energy fluoride and chloro-fluoride glasses, J. Non-Cryst.
Solids, 213–214, pp. 30–35.
15. Matecki, M., Poulain, Mi., Poulain, Ma. (1983). Cadmium halide glasses,
J. Non-Cryst. Solids, 56(1–3), pp 81–86.
References 333
16. Poulain, Mi., Matecki, J. M., Mouric, L., Poulain, Ma. (1983). Cadmium
halide glasses II. Chloride glass, Mater. Res. Bull., 18(5), pp. 631–636.
17. Matecki, M., Poulain, Ma. (1992). Composition adjustments in
fluorochloride glasses, J. Non-Cryst. Solids, 140, pp. 82–86.
18. Poulain, Mi., Poulain, Ma. Matecki, M. (1986). Rare-earths in fluoride
glasses: LnF3, AlF3, ThaF4, BaF2 systems and related glasses, J. Non-
Cryst. Solids, 51(2), pp. 201–215.
19. Bouaggad, A. (1986). PhD thesis, Rennes University, France.
20. Lecoq, A., Poulain, Ma. (1980). Etude phenomenologique du role
stabilisateur de l’aluminium dans les verres au tetrafluorure de
zirconium, Verres Refct., 34(3), pp. 333–342.
21. Jacoboni, C., Boulard, B., Baniel, P., Poignant, H. (1987). Vapor
deposition of fluoride glasses, Mater. Sci. Forum, 19–20, pp. 253–258.
22. Ohishi, Y., Mitachi, S., Kanamori T., Manabe, T. (1983). Optical absorption
of 3d transition metal and rare-earth elements in zirconium fluoride
glasses, Phys. Chem. Glasses, 24, pp. 135–140.
23. France, P. W. (1987). PhD thesis, Chapter 6, Sheffield University.
24. France, P. W., Carter, S. F., Williams, J. R., Beales, K. J., Parker, J. M. (1984).
OH-absorption in fluoride glass infrared fibers, Electron. Lett., 20(14),
pp. 607–608.
25. Drexhage, M. G. (1985). Preparation and properties of high quality
bulk fluoride glasses, Mat. Sci. Forum, 5, pp. 1–18.
26. Fujiura, K., Ohishi, Y., Sakaguchi, S., Ternuma, Y. (1989). Synthesis of
high-purity zirconium fluoride for fluoride glass fibers by chemical
vapor deposition, J. Am. Ceram. Soc., 72(8), pp. 1368–1372.
27. Withers, H. P. Jr., Monk, V. A., Cooper, G. A. (1989). High purity zirconium
tetra-fluoride for fluoride glass applications, Proc. SPIE, 1048, pp. 72–
77.
28. Saad, M. (2009). High purity fluoride glass synthesis: a review, Proc.
SPIE, 7228, p. 72280G.
29. Poulain, M. (1991). Chapter 1 Fluoride glass compositions and
processing, in Fluoride Glass Fiber Optics, Aggarwal, I., Lu, G., eds.
(Academic Press, Boston), pp. 1–35.
30. Fonteneau, G., Tregoat D., Lucas, J. (1985). Determination du Coefficient
d’Extinction de OH- dans les verres fluores a base de metaux lourds,
Mater. Res. Bull., 20(9), pp. 1047–1051.
31. Nakai, T., Mimura, Y., Tokiwa, H., Shinmori, O. (1986). Dehydration of
fluoride glasses by NF3 processing, J. Lightwave Technol., LT-4, 1.
334 Fluoride Glasses and Fibers
47. Aggarwal, I. D., Lu, G. (1991). Fluoride Glass Fiber Optics (Academic
Press, Harcourt Brace Jovanovich, San Diego, CA).
48. Carter, S. F. (1990). Mechanical properties, in Fluoride Glass Optical
Fibers, P. W. France, ed., Blackie USA and Canada (CRC Press), p. 219.
49. Poulain, Ma., Gougeon, N., Maze, G. (2003). Mechanical properties of
fluoride glass fibers, Proc. SPIE, 4940, pp. 30–46.
50. Xiushan, Z., Peyghambarian, N. (2010). High-power ZBLAN glass fiber
lasers: review and prospect, Adv. OptoElectron., 2010, Article ID.
501956, 23 pages.
51. Fortin, V., Bernier, M., Bah, S. T., Vallee, R. (2015). 30W fluoride glass
all-fiber laser at 2.94 µm, Opt. Lett., 40(12), pp. 2882–2885.
52. Quimby, R., Saad, M. (2013). Dy-fluoroindate fiber laser at 4.5 μm with
cascade lasing, Adv. Solid State Lasers, paper AM2A.7.
53. Epstein, R. I., Buchwald, M. I., Edwards, B. C., Gosnell, T. R., Mungan, C.
E. (1995). Observation of laser-induced fluorescent cooling of a solid,
Nature, 377, pp. 500–503.
54. Rayner, A., Friese, M. E. J., Truscott, A. G., Heckenberg, N. R., Rubinsztein-
Dunlop, H. (2001). Laser cooling of a solid from ambient temperature,
J. Mod. Opt., 48(1), pp. 103–114.
55. Murtagh, M. T., Sigel Jr., G. H., Fajardo, J. C., Edwards, B. C., Epstein,
R. I. (1999). Laser-induced fluorescent cooling of rare earth–doped
fluoride glasses, J. Non-Cryst. Solids, 253, pp. 50–57.
56. Hoyt, C. W., Hasselbeck, M. P., Sheik-Bahae, M. (2003). Advances in
laser cooling of thulium-doped glass, J. Opt. Soc. Am. B, 20(5), pp.
1066–1074.
Chapter 11
11.1 Introduction
In the solid phase thermal energy is largely contained in the
vibrational modes of the lattice. Upon interaction with laser light,
heat can be removed from the system through annihilation of lattice
phonons.
Optical cooling in solids is based on the principle of anti-Stokes
luminescence, where coherent excitation below the mean emission
energy of the electronic transition induces absorption of energy from
the lattice, and hence annihilation of lattice phonons, via ultrafast
thermalization of the electronic excitations (Fig. 11.1). Absorption of
photons at one frequency is followed by spontaneous re-emission of
blue-shifted photons. The extra energy is removed from the lattice,
resulting in bulk net cooling.
Introduction 339
λ>λf
λf
Figure 11.4 Schematic (a) and picture (b) of the Czochralski furnace at the
NMLA Laboratory of Pisa University.
(a) (b)
Figure 11.5 (a) Interior of the growth chamber. The heater element, with
cylindrical geometry, accommodates the crucible holder in the
middle. Beneath the resistance are placed the two electrodes.
(b) Platinum crucible used for the growth of Yb-doped YLF
single crystal.
the bottom end of the rod to move back and forth from the secondary
chamber to the bottom of the crucible. Minimum displacements
achievable are of the order of the tenth of a millimeter. The slowest
translational speed is 0.1 mm/h. The stepper motor that produces
the revolution of the shaft enables rotation rates of the order of a few
revolutions/min, normally required to attain thermal symmetry.
The puller shaft is terminated by a dedicated extension (seed
holder), which consists of a cylindrical tube, made of high-purity
platinum (99.95%). On the bottom end of the seed holder extension,
crystal seeds and the cleaning tools are fastened by using high-purity
platinum wires (99.95%).
The diameter of the growing crystal is controlled by using an
optical system computer controlled (see Fig. 11.6).
an asymmetric spot at the interface so that the laser beam can shine
the meniscus while it is moving toward the crucible edge when
the crystal diameter increases. The reflection position acquired in
a revolution period is converted in a diameter value by means of
a suitable algorithm. The laser and the CCD camera are arranged
inside a compact support placed outside the growth chamber, in
proximity of one of the top window of the main body, through which
the incident beam is passed and the reflection from the meniscus
is acquired. The diameter control is operated by means of a second
PID feedback loop, which automatically varies the current supplied
to the resistance, and hence the melt temperature, according to the
difference between measured and set values of the crystal diameter.
Diameter fluctuations attainable in growth conditions are of the
order of 1%.
Table 11.1 Main physical and optical properties of the YLF crystal
In case of co-doping with two different species of RE3+ ions, the YF3
compound needs to be replaced with appropriate amounts of both
the RE compounds. The dopant RE3+ ions, present in the melt, will be
incorporated in the crystal while it is growing, replacing the Y3+ host
ions in the unit cells.
Liquid
YF3 + L
T [K]
A B
Figure 11.8 Phase diagram of the LiF-YF3 system. The label A indicates a
phase with solid LiF mixed with liquid. The label B indicates a
phase with solid YLF and liquid [19].
transparent. In Fig. 11.10b it can be seen the crystal seed with the Pt
wire, used to fasten the crystal seed to the seed holder extension, still
attached to the crystal boule. In both Fig. 11.10c and Fig. 11.10d is
visible the narrow crystal neck operated in the early stage of growth
process by changing the melt temperature. Narrow diameters of 1–3
mm (Fig. 11.10c) and 2–3 mm (Fig. 11.10d) were grown for up to
10 mm (Fig. 11.10c). The lateral crystals edges and the elliptic cross
section can be observed in Fig. 11.10a,c,d. In Fig. 11.10e the circular
cross section is indication of a c axis component along the growth
axis. Irregular diameter changes as those visible in Fig. 11.10a,b,d
did not reflect in the inclusion of defects. An example of excellent
diameter stability is provided in Fig. 11.10c.
867.5
867.0
866.5
Temperature [°C]
866.0
865.5
865.0
864.5
20 40 60 80 100 120 140 160 180
Time [h]
14
12
10
Diameter [mm]
0 20 40 60 80
Length [mm]
(b) (c)
(a)
(d) (e)
Figure 11.10 As-grown crystals of YLF:5%Yb (a and b), YLF:7.5% (d), and
YLF:10%Yb (c and e).
(a) (b)
Figure 11.11 (a) Yb2+ contamination of Yb-doped YLF single crystals grown
under a CF4-depleted atmosphere. (b) Carbon deposition
on crystal surface and furnace interior due to excessive
concentration of CF4 in the growth atmosphere.
364 Crystal Growth of Fluoride Single Crystals for Optical Refrigeration
(a) (b)
Figure 11.12 (a) As-grown crystal boule of YLF undoped crystal. (b) Seeds of
YLF undoped crystal oriented along the a axis.
Due to the similarity of Yb3+ and Tm3+ ionic radii sizes, a segregation
coefficient equal to 1 is expected, even for the substitution of Y3+/
Tm3+ substitution [11].
The growth process was carried out under an argon and CF4
atmosphere. During the growth, the pulling rate was 0.5 mm/h and
the rotation speed 5 rpm. As a crystal seed, a YLF undoped single
crystal oriented along the a axis was used.
Due to the low concentrations of the TmF3 compound required for
the desired doping level, the actual concentration of Tm was checked
after the growth with a trace element analysis performed by means
of laser ablation–inductively coupled plasma–mass spectrometry
(LA-IPC-MS) [20]. A Tm concentration of 16 ppm was detected. No
internal defects related to the Yb–Tm co-doping were detected.
cross section of the sample. The growth axis resulted in the plane c-a,
tilted about 30° with respect to the a axis. In Fig. 11.14 is reported
the diffraction pattern acquired for the as-grown crystal boule of YLF
undoped, by shining the X-ray beam perpendicularly to the growth
axis.
a
a c
a
c a (a) c a (b)
a c
growth axis for the self-nucleation in YLF crystal is the c axis. The
measurement of the crystallographic axes orientation returned the c
axis tilted just a few degrees with respect to the growth axis.
(d) (e) (f )
Figure 11.15 Test of crystal quality with 633 laser beam. (a) Single scattering
center detected inside a cooling sample; (b) scattering-
free area of the same sample; (c) high-density aggregate of
scattering centers; (d) propagation of 633 nm laser beam
in the neck of a crystalline boule: high density of scattering
defects; (e) middle region: less numerous scattering center;
and (f) bottom: scattering almost absent.
(a) (b)
Figure 11.17 (a) Layout of right angle–cut sample prepared for cooling
experiments in single-pass configuration. (b) Picture of
finished cooling sample of YLF:10%Yb.
Acknowledgments
The authors would like to acknowledge I. Grassini for her competence
and care in preparing the samples and D. Parisi and S. Veronesi for
useful discussions and their expertise in crystal growth. A. Volpi
acknowledges support by the European Space Agency under grant
no. 4000108074/13/NL/PA, “Cooling Effect on Fluoride Crystals.”
References 375
References
12.1 Introduction
The conceptual foundation of optical refrigeration of solids was laid
by Pringsheim [1]. Soon after the first major experimental break-
through by Epstein et al. [2], who achieved net cooling of an Yb-
doped ZBLANP glass, the study of the physics and applications of
optical refrigeration of solids became a major worldwide research
area, as documented by review articles, including those written by
Sheik-Bahae and Epstein [3] and Nemova and Kashyap [4]. Works
following the 1995 break-through also included research on the
cooling of inorganic semiconductors, which are generally believed
to hold promise for cooling to temperatures of about 10 K. The
semiconductor research includes various theoretical aspects, for
example those discussed by Oraevsky [5], Rivlin and Zadernovsky
[6], Sheik-Bahae and Epstein [7], Apostolova [8], and Khurgin [9],
as well as experimental studies, including, for example, the ones
presented by Gauck et al. [10] and Imangholi et al. [11]. Important
recent milestones are the observation of cooling to cryogenic
temperatures of Yb-doped YLF crystals reported by Seletskiy et al.
[12] and the cooling of nanostructured CdS crystals reported by
Zhang et al. [13].
Inorganic semiconductors, such as GaAs or CdS crystals, are
commonly characterized by their electronic band structure. In the
vicinity of the fundamental bandgap, the band structure can be
approximated by two bands, the conduction and the valence band
(Fig. 12.1). However, optical transitions involved in absorption and
photoluminescence processes are generally not well described by
band-to-band transitions, owing to the strong Coulomb interaction
between electrons in the conduction band and holes in the valence
bands. The Coulomb interaction can give rise to sharp optical
resonances called excitonic resonances, which correspond to
the bound electron–hole pairs (excitons). Optical refrigeration
of semiconductors is dominated by excitonic resonances, both
in the absorption (pumping) and the upconverted (anti-Stokes)
luminescence process. While at low excitation densities (low
pumping intensities) the excitons can be described with simple
Bosonic models, such descriptions are not appropriate for optical
refrigeration, as the strong pumping gives rise to relatively large
excitation densities and hence to the partial ionization of excitons.
Theoretical Foundation and the Importance of Excitonic Effects 379
a (w ) =
4pw
n bc
Im (c R
) (12.1)
(q = 0, w )
Ir I t
AQW (w ) ∫ 1 - - ª Ims R (w ) (12.5)
I0 I0
where I0, Ir, and It are the incident, reflected, and transmitted
intensities, respectively.
ε
c
k
v
sjk
where
djs(k) (12.8)
382 Microscopic Theory of Optical Refrigeration of Semiconductors
denotes the interband dipole matrix element (for details of this and
related matrix elements see Gu et al. [19]). The imaginary and real
parts of the free two-particle Green’s function are given by
{ }
sj (k , q , w ) = -
Im g0R
1 dw ¢
2 2p Ú { }
1 - f s (w - w ¢ ) - f j (w ¢ )
(12.9)
¥ As (k s ,w - w ¢ )A j (k j ,w ¢ )
and
{ } Ú { }
dw ¢ Im gsj (k , q ,w ¢ )
0R
(12.10)
sj (k , q , w ) = P
Re g0R
p w¢ - w
where
1 (12.11)
fa (w ) = ( w - ma )/kBT
e +1
are the Fermi functions for electrons (a = e) and holes (a = h), and the
spectral functions are given by
{
Aa (k ,w ) = -2Im GaR (k ,w ) (12.12) }
In its present stage, our theory does not include a microscopic
description of phonons and electron–phonon coupling. However,
for optical refrigeration it is important to account for excitonic line-
broadening effects since the optical pump beam is usually chosen
to be in resonance with the low-frequency tail of the lowest exciton
resonance. We therefore include a phenomenological amendment to
the spectral function that accounts for such line-broadening effects.
The temperature dependence of the line broadening due to coupling
of electrons to longitudinal acoustic (LA) and longitudinal optic (LO)
phonons is approximately given by
g LO
g ph
0
(T ) = g LAT + WLO / kBT
(12.13)
e -1
This yields the following phonon-broadened spectral functions
for electrons and holes:
dw ¢
Ú 2p
2
Aaph (k ,w ) = Aa (k ,w - w ¢ ) (12.14)
pg ph
0
cos h(2w ¢/g ph
0
)
The retarded (superscript R) single-particle Green’s function is
given by
Theoretical Foundation and the Importance of Excitonic Effects 383
GaR (k ,w ) =
w - e a (k ) - S aR,T (k ,w ) - S R,exc
a (k ) - S R,CH
a (k )
(12.15)
where the unrenormalized single-particle energies (i.e., the
parabolic band structure, parametrized in terms of effective masses
ma) is given by
2
2 k a
e a (k a ) = (12.16)
2ma
and the real and imaginary parts of the T matrix (superscript T) self-
energies (see Kwong et al. [14] and also Schmielau et al. [20] and
Piermarocchi and Tassone [21]) are
Re {
S Ra ,T (k ,w )} Ú =P
R ,T
{
dw ' Im S a (k ,w ¢ )
(12.17)
}
p w¢ - w
and
{
Im S R,T
a (k , w ) = } Â Ú d2wp¢ {g(w + w ¢) + f (w ¢)}
a¢
a¢ k ¢ (12.18)
¥ Im {T ¢¢ (p , p , q ,w + w ¢ )} A (k ¢ ,w ¢ )
R
aa¢ a¢
where q = k + k¢ and
ma¢ k - ma k ¢
p= (12.19)
ma + ma¢
Furthermore, the real-valued exchange (superscript “exc”) self-
energy is given by
dw
S R,exc
a (k ) = - ÂW
k¢
aa (k - k¢ )Ú 2p f (w )A (k ,w ) (12.20)
a a
ÂW
k ¢¢
aa ¢ (k - k ¢¢ , w ) ga0R (k ¢¢ , q , w )Taa
R
¢ (k ¢¢ , k ¢ , q , w )
(12.23)
4p e2 (12.24)
Waa ¢ (q ) = za za ¢
e b (q2 + k (q)2 )
where e b = nb2 , κ(q) is the inverse screening length, and za the sign
for the charge (–1 for electrons and +1 for holes). The unscreened
potential is V(q) = 4πe2/(εbq2). Details on how the effective potential
can be treated in a numerical solution of the equation set can be
found in Kwong et al. [14]. Note that the above set of equations
constitutes a self-consistency problem, as the self-energies depend
on the spectral function, but the spectral functions in turn depend
on the self-energies.
Once the luminescence spectral density is known, we can use the
cooling model developed by Sheik-Bahae and Epstein [7]. We define
the radiative recombination coefficient B (for ne = nh ∫ n ) via
dw
L(n) = B(n)n2 = Ú 2p R(w ) (12.25)
and use the rate equation for the excitation density evaluated under
stationary conditions,
dn a(wa , n)
= I - An - heL(n) - Cn3 = 0 (12.26)
dt wa pump
where Ipump is the pump intensity (more precisely irradiance), ωa
the frequency of the pump field, A the nonradiative recombination
coefficient, he the extraction efficiency of the luminescence (i.e.,
the fraction of the luminescence that leaves the crystal), and C the
Auger coefficient. Equation 12.26, together with the power balance
equation for the net power extracted from the system,
w =
Ú dw w R(w ) (12.28)
Ú dw R(w )
the mean luminescence frequency, constitutes the basic framework
of the Sheik-Bahae–Epstein model.
The break-even nonradiative recombination coefficient Ab, which
marks the boundary where cooling becomes possible, is defined
by Pnet(Ab) = 0. Fixing all but two parameters, leaving only the
experimentally easily adjustable parameters ωa and ne as variables,
we optimize the break-even condition in the two-dimensional space
of ωa and ne, resulting in the optimized break-even nonradiative
coefficient
{
A ∫ max A(w a , ne ) P
net =0
} (12.29)
Figure 12.2 shows the results for the optimized breakeven non-
radiative lifetime (defined as the inverse of *A) as a function of
temperature. The solid and dotted lines show, respectively, the results
of the full calculation and the second-order approximation using the
free single-particle Green’s function G0 in the T matrix self-energies.
The overall temperature dependence is similar for these two curves
(the luminescence spectra in the full T matrix and second-order
approximations are quite different, but the resulting differences
in the values of the optimal density and radiative recombination
coefficient largely compensate each other, leading to similar values
for *τb), suggesting that many-body correlation effects beyond the
T matrix approximation would probably not change the results
significantly. In the so-called plasma theory model, the T matrix
self-energy is completely neglected, but the excitonic effects in the
susceptibility are still taken into account at a nonperturbative level.
In other words, in that model the exciton resonance is accounted
for in the susceptibility, but the many-particle processes that
renormalize the exciton resonance are reduced to pure electron–
hole plasma effects. We see that, qualitatively, plasma theory yields
results similar to the full theory. Only the model where all Coulomb
interactions are neglected, called “free” model, yields qualitatively
different results. We see from Fig. 12.2 that the existence of the
exciton resonance is the main factor that can make cooling at low
temperatures possible. More details can be found in Refs. [14–16].
386 Microscopic Theory of Optical Refrigeration of Semiconductors
Full
1000
2nd order
free
100 Plasma
10
(µs)
1
τb
0.1 Excitonic
Models
0.01
10 100 500
Temperature (K)
Figure 12.2 Optimized break-even nonradiative lifetime versus tempera-
ture for several different theories. For each model, the predict-
ed cooling region is above its curve. Reprinted from Ref. [16],
Copyright 2007, by the American Physical Society.
z
air
GaAs J q σ ω ( r ) d
ZnSe
Figure 12.3 Schematic of the geometry analyzed with the spectral energy
flux density. In the calculations, we assumed the lateral
extension to be infinite and as a consequence the net energy
flow to be restricted to positive and negative z directions
(thick blue vertical arrows). The extension in the z direction
was chosen to be 1 μm, and the partial or total reflection at the
GaAs/air and GaAs/ZnSe interfaces were taken into account
according to Fresnel’s equations.
In the last step, we defined the spectral density of the energy flux
Jqσω(r, t), which obeys the following transport equation:
Ï∂
Ó ∂ t
R ¸
( ) <
Ì + v q w q ◊ — - Im Pq w ˝ J q s w (r , t ) = - q Im Dq w i Pq w
˛
R
( )
(12.32)
Here, the second term in the parenthesis on the left-hand side
accounts for spatial propagation in the q direction with generalized
group velocity vqω, the third term accounts for luminescence re-
388 Microscopic Theory of Optical Refrigeration of Semiconductors
 ÂÚ
1 1
he B(n) n2 = dw z s ◊ J q s w ( zsurface ) t q s w ( zsurface )
dz q s
w q
(12.33)
surface
 ÂÚ
1
dw z s ◊ J q s w ( zsurface ) t q s w ( zsurface )
w l he B(n) n2 =
dz qs
(12.34)
surface
e-h polariton
w
2s
1s
q
Figure 12.4 Exciton–polariton dispersion at a temperature of 5 K, as seen
in the peaks of the logarithm of the photon spectral function
[-2ImDqw R
]. The anticrossing between the 1s exciton and the
photon dispersion is clearly visible, and to a lesser extent even
that of the 2s exciton and the photons. In the e–h continuum,
the polariton dispersion is dominated by that of free photons.
The (quadratic) spatial dispersion of the 1s and 2s excitons
cannot be seen in this figure because the chosen q range is very
small.
1010
Luminescence
106
102
-1.0 -0.5 0.0
(ω- ER) / Eg
Figure 12.5 Photoluminescence spectra at T = 5 K as hypothetically seen
inside the semiconductor with (red curve) and without (blue
curve) polariton effects. The polariton luminescence contains
a large component from the lower polariton branch (at large
wavevectors), which however are not transmitted to the
outside (see Fig. 12.6).
1010
Luminescence
106
102
-1.0 -0.5 0.0
(ω - ER)/Eg
Figure 12.6 Photoluminescence spectra at T = 5 K as seen outside the
semiconductor with (red solid curve) and without (blue dashed
curve) polariton effects (these two curves are practically
indistinguishable). The green curve shows results neglecting
luminescence re-absorption.
Effect of Luminescence Propagation and Re-Absorption 391
Loutside
ηe =
0.15 Llocal
ηe
0.10
N a = 3 ×1015 cm −3
0.05
10-4
N a = 3 ×1015 cm −3
ZnSe dome structure
10-5
Lifetime (s)
t t = 8µs
-6
10
10-7 he t
10-8
0 100 200 300 400 500
Temperature (K)
Figure 12.8 Calculated minority carrier lifetime for a p-doped GaAs
structure with a ZnSe dome.
light
beam
r2
−
I pump (r ) = I 0 e 2σ 2
spatial profile
of excitation
Lout (r ) = ηe L(r )
Lin (r ) = (1 − ηe ) L(r )
out in
r
0
Figure 12.10 (Top sketch) Sketch of luminescence propagation. A fraction
ηe of the luminescence produced at a given position r in the
sample escapes from the sample, whereas the rest (1 – ηe)
propagates laterally in the sample and is re-absorbed for
sufficiently laterally extended samples. (Bottom sketch) In our
model, the re-absorption at position r has two contributions
from lateral photon propagation, one from luminescence
propagation laterally outward and the other laterally inward.
Reprinted with permission from Ref. [24], Copyright 2009,
Society of Photo Optical Instrumentation Engineers.
(12.37)
(where fq is the azimuthal angle of q). The transport equation,
including the luminescence source term and the luminescence re-
absorption term, with α(r, ω) ∫ α (n(r), ω), reads
Finite Spatial Beam Profiles 395
Ï 1 ∂ ¸
̱ r + a (r ,w )˝ Jqout/in
,s ,w (r ) = sq,w ,s (r ) (12.38)
Ó r ∂ r ˛
where the upper sign is for Jout and the lower sign for Jin, respectively.
Using the boundary conditions Jq,s ,w ( • ) = 0 and
in
Jq,s ,w (0) = Jq,s ,w (0) (which accounts for the fact that inward flow
out in
• Ê r
- Ú dr ¢¢a ( r ¢¢ ,w )
r
- Ú dr ¢¢a (|r ¢¢|,w )
ˆ
Á ˜
Ú
1
Jqs w (r ) =
out
dr ¢r ¢ s(r ¢ , q ,w )Á q(r - r ¢ )e r ¢ + e -r ¢ ˜
r
0 ÁË ˜¯
(12.39)
and the incoming flux is
r¢
•
Ú
- dr ¢¢a ( r ¢¢ , w )
Ú
in 1
Jqs w (r ) = dr ¢r ¢ s(r ¢ , q , w ) e (12.40)
r
r
r
The luminescence source sq,ω,σ(r) is normalized such that the total
in-plane (lateral) luminescence (i.e., sq,ω,σ(r) summed/integrated
over all wavevectors, frequencies, and polarization states) yields the
fraction 1 – he of the globally generated luminescence (integrated
over all space). We have modified the rate equations entering
the Sheik-Bahae–Epstein model by including the re-absorption
calculated with the lateral photon transport model. The modified
rate equation for the density profile reads
a (w a , n(r ))
Ipump (r ) - An(r ) - B (n(r )) n(r )2 - Cn(r )3 + Qn (r )
w a
dn(r )
= =0
dt
(12.41)
Note that the luminescence term (Bn2) is not multiplied by the
extraction efficiency. Instead, we include a source term that accounts
for the re-absorption of carriers due to lateral luminescence
propagation, which is given in terms of the photon flux and the
absorption coefficient,
Qn (r ) = Â a (r ,w ) | J
q,s ,w
q,s ,w (r ) | (12.42)
396 Microscopic Theory of Optical Refrigeration of Semiconductors
Qe (r ) = Â a(r,w )w | J
q,s ,w
q,s ,w (r ) | (12.44)
Figures 12.11 and 12.12 show results with and without the lateral
photon transport included. We see that, apart from slight numerical
noise, the results for the power is very similar in both cases,
implying that lateral photon transport under typical experimental
conditions should not be expected to prohibit cooling. As expected,
the density profile in the case with lateral photon transport is
smeared out beyond the region of optical excitation by the external
light beam. However, the off-axis density produced by luminescence
re-absorption is much smaller than the density excited directly by
the external beam. This explains why the power profiles are very
similar in the two cases (with and without lateral photon transport).
1016
1015
1014
0 100 200 300 400 500
Distance from center (µm)
Figure 12.11 Spatial density profile for excitation with a Gaussian beam
(σ = 100 μm) at temperature T = 300 K with spatial photon
transport. Reprinted with permission from Ref. [24], Copyright
2009, Society of Photo Optical Instrumentation Engineers.
Theory of Passivation Layers 397
2x105
-6x105
0 100 200 300 400 500
Distance from center (µm)
Figure 12.12 Spatial profile of the net power for excitation with a Gaussian
beam (σ = 100 μm) at temperature T = 300 K with spatial
photon transport. Reprinted with permission from Ref. [24],
Copyright 2009, Society of Photo Optical Instrumentation
Engineers.
and holes. It had been known for some time that the band profile of
this structure is of type II character, which means that the conduction
band has a maximum in the GaAs region. In that case, the electrons
tend to accumulate in the energetically more favorable GaInP layers,
away from the holes that remain in the GaAs layer. Such a spatial
charge separation would hinder photo-luminescence and hence
could be detrimental for optical refrigeration. To address this issue,
we first determine the spatial band profile and electron and hole
density profiles across the n-p-n structures. With that knowledge,
we use our microscopic photoluminescence and absorption theory
to determine the spatial variation in the luminescence efficiency and
the absorption spectra.
n p n
Figure 12.13 Schematic of the doping profile of the structure, which was used
in the University of New Mexico experiments, see Imangholi et
al. [11], with typical doping concentrations.
ne ( z ) = 2Â
1
(12.47)
1+e
k
(
b e e (k )- eV ( z )- me - e c0 ( z ) )
and similarly for the valence band holes,
nh ( z ) = 2 Â
1
(12.48)
b (e (k )+ eV ( z )- mh - e v0 ( z ))
1+e h
k
Ú dz n charge ( z ) = 0 (12.52)
Initialize V(z)
1.0
1.4224 eV
1.91 eV
0.5
0.0
-0.5
Valence band
1.5
1.0
Energy (eV)
0.5
0.0
-0.5
- - - - - - - - - - - - - - - - - - -
n p n
+ + ++++ ++++ + +
zc zj ze
Figure 12.17 Sketch of n-p-n structure with electrons and holes schematically
indicated.
∂neopt ( z ) ∂nhopt ( z )
∂t
=
∂t
=- Â
j =1 ,2
d ( z - z j ) S nmin ( z j ) (12.56)
nr nr
1016
Density (cm-3)
1014
1010
-1.0 -0.5 0.0 0.5 1.0
Distance from center (µm)
Figure 12.18 Electron (solid green line) and hole (dashed red line) density
profile at T = 300 K and with an acceptor density of Na = 3.5
× 1015 cm–3. The optical excitation is chosen to yield a carrier
density at the center of 1015 cm–3. Holes are majority carriers
at the center but minority carriers at the interface.
1
na = Ú dz n a ( z ) (12.57)
opt
d
and assume the layer thickness to be sufficiently large so that the
average density is the same as the density at the center of the layer,
na ª naopt ( zc ) for a = e or h, we have
∂neopt ( z c ) ∂nhopt ( z c ) 2S
= =- n (z )
∂t ∂t d min j
nr nr
2S nmin ( z j )
=- n ( z ) (12.58)
d nmin ( zc ) min c
= - A nmin ( zc )
404 Microscopic Theory of Optical Refrigeration of Semiconductors
100
nh(zj) / nmin(zc)
10-2
10-4
10-6 10
10 1012 1014 1016 1018
nmin(zc)
Figure 12.19 Dependence of the hole density at the interface versus minority
center density. The solid line is for Na = 1014 cm–3, Nd = 1017
cm–3, the dash-dotted line for Na = 1014 cm–3, Nd = 1019 cm–3,
and the dashed line for Na = 3.5 × 1017 cm–3, Nd = 1017 cm–3. The
thickness d is taken to be 1 μm and the recombination velocity
S = 500 cm/s.
hole density (Fig. 12.19) and hence improve (increase) the effective
nonradiative lifetime (Fig. 12.20). As shown in Fig. 12.20, such an
increased n-doping concentration can, for the idealized model of
and infinitely thin interface region, yield lifetimes comparable to
the optimized break-even lifetimes discussed above. This, however,
assumes that a high n-doping concentration does not open up new
nonradiative transitions that would again decrease the effective
nonradiative lifetime.
106
n
104
(µs)
102
τnr
100 τb
10-2 10
10 1012 1014 1016 1018
nmin(zc)
Figure 12.20 Effective nonradiative recombination time as a function of
center density. The three different thick line styles correspond
to the same doping levels as in Fig. 12.19. The vertical and
horizontal dotted lines indicate the optimized cooling density
and optimized break-even lifetime, respectively, obtained from
the model of bulk semiconductors discussed above in Section
12.2.
0.0
ηc
-0.1
-0.2 10
10 1013 1016 1019
Minority center density (cm –3)
Figure 12.21 Cooling efficiency versus minority center density. The three
different line styles correspond to the same doping levels as
in Fig. 12.19. The dash-dotted green line shows that cooling
should be possible, hc > 0 , for low to moderate minority center
densities (i.e., moderate optical pumping powers) and large
n-doping concentration in the passivation layers. Reprinted
with permission from Ref. [23], Copyright 2010, AIP Publishing
LLC.
Acknowledgments
We thank M. Sheik-Bahae, B. Imangholi, M. Hasselbeck, R. Epstein,
and J. Khurgin for helpful discussions. We acknowledge financial
support from AFOSR MURI grant no. A9550-04-1-0356 and addi-
tional support from JSOP and TRIF Photonics.
References
19. Gu, B., Kwong, N. H., Binder, R. (2013). Relation between the interband
dipole and momentum matrix elements in semiconductors, Phys. Rev.
B, 87, p. 125301.
20. Schmielau, T., Manske, G., Tamme, D., Henneberger, K. (2000).
T-matrix approach to the linear optical response of highly excited
semiconductors, Phys. Status Solidi B, 221, pp. 215–219.
21. Piermarocchi, C., Tassone, F. (2001). Role of bound pairs in the optical
properties of highly excited semiconductors: a self-consistent ladder
approximation approach, Phys. Rev. B, 63, p. 245308.
22. Rupper, G., Kwong, N. H., Binder, R. (2008). Theory of luminescence
and optical refrigeration in p-doped semiconductors, Proc. SPIE, 6907,
p. 690705.
23. Rupper, G., Kwong, N. H., Binder, R., Li, Ch.-Y., Sheik-Bahae, M. (2010).
Effect of n-p-n heterostructures on interface recombination and
semiconductor laser cooling, J. Appl. Phys., 108, p. 113118.
24. Rupper, G., Kwong, N. H., Binder, R. (2009). The role of finite spatial
beam profiles on photo-luminescence and laser cooling in GaAs
structures, Proc. SPIE, 7228, p. 722805-1.
25. Rupper, G., Kwong, N. H., Binder, R. (2010). Theory of time-resolved
photo-luminescence and carrier lifetime measurements in GaAs/
GaInP heterostructures, Proc. SPIE, 7614, p. 76140D-1.
Chapter 13
13.1 Introduction
Cooling materials with light is a topic that has attracted tremendous
interest in recent years. Researchers usually find the theory
counterintuitive, and, in fact, it was a highly disputed idea when
Pringsheim first proposed it in 1929 [35]. However, only 66 years
later, the first observation of laser cooling of solids was made by
Epstein in 1995 in trivalent ytterbium-doped heavy metal fluoride
glass [12]. In rare earth ions, such as ytterbium, the electronic 4f
levels are shielded from their surroundings by the filled 5s and 5p
shells, leading to suppression of multiphonon relaxation and a high
radiative recombination efficiency favorable for optical refrigeration
[36]. Since the first discovery of laser cooling, a methodical
kT
Input Laser
Photo -
Luminescence
kT
Figure 13.1 The process of anti-Stokes laser cooling of semiconductors.
Introduction 411
where Epl is the average energy of the PL, Ei is the average energy
of the incident photons, and ηex and ηabs are the external quantum
efficiency and the absorption efficiency, respectively. The external
quantum efficiency (ηex) can be calculated as [42]:
he Bn2
hex = (13.2)
An + he Bn2 + Cn3
where ηe is the extraction efficiency [28], n is the photogenerated
carrier density, and A, B, and C are the SRH, radiative, and Auger
recombination coefficients, respectively. This parameter is a measure
of the fraction of luminescence power extracted out of the material
before getting reabsorbed. The absorption efficiency is equal to:
a
habs = (13.3)
a + ab
where α is the absorption in the active region and αb is the background
absorption due to free carriers, impurities, and the defects of the
material. It is straightforward to conclude from Eqs. 13.2 and
13.3 that in order to maximize the net cooling efficiency, both the
external quantum efficiency and the absorption efficiency should
be maximized. The external quantum efficiency is dependent on the
material quality, the extraction efficiency, and the carrier density.
The extraction efficiency can be increased to almost 100% using
novel techniques that suppress both the total internal reflection
and the Fresnel reflection inside a high-index material. However,
the material quality sets a limit on maximum attainable external
quantum efficiency and absorption efficiency. The materials of choice
for semiconductor laser cooling up to now are GaAs [42] and CdS
[51]. Considering the bandgaps of these materials, the maximum net
cooling efficiency is approximately 2% for gallium arsenide and 1%
for cadmium sulfide at room temperature. These values are quite
low, rendering the laser cooling of bulk GaAs impossible, and for CdS,
only a particular nanoribbon structure has been cooled using a laser.
This is due to the insufficient external and absorption efficiencies
of these materials. These challenges suggest that mechanisms other
than anti-Stokes should be developed and employed in order to have
practical laser cryocoolers in the future. While different methods
of laser cooling for gaseous systems (such as Doppler cooling,
Sisyphus, resolved-sideband cooling, and sympathetic cooling) have
Piezoelectricity in Semiconductors 413
∂ua
a , b = , u = u,x u y , uz x b = x , y , z (13.6)
∂x b a
where uα is the lattice displacement along the direction specified by
xα. The strain tensor is symmetric and the following relation holds
[50]:
∂ua ∂ub
ab = ba = + (13.7)
∂x b ∂xa
Equations 13.4–13.7 provide a complete description of
piezoelectric field as a result of different types of stress and strain in
the material. It is important to note that because of crystal symmetry,
many elements of the elastic constants and the piezoelectric tensor
are essentially zero. For example, for a cubic crystal, the values of the
uniaxial elastic constants along the three axis are the same because
of spatial invariance (Cii=Cjj, i ≠ j). A similar relation holds for shear
components. Furthermore, shear strain cannot produce a normal
stress (e.g., C41 = 0) , and as a result the elasticity tensor is simplified
to [32]
Ê C11 C12 C12 0 0 0 ˆ
ÁC C11 C12 0 0 0 ˜
Á 12 ˜
ÁC C12 C11 0 0 0 ˜
Ccubic = Á 12 (13.8)
Á 0 0 0 C 44 0 0 ˜˜
Á 0 0 0 0 C 44 0 ˜
Á ˜
Ë 0 0 0 0 0 C 44 ¯
The above matrices are used to find the stress in materials grown
along (0 0 1) and (0 0 0 1) for the case of cubic and hexagonally
Piezoelectricity in Semiconductors 415
Movable mirror
x
Optical cavity
Figure 13.3 Schematics of the system suitable for optomechanical
backaction interaction. It consists of a movable mirror (the
displacement of the mirror is shown by x) and resonating
optical field that is coupled to the mechanical state of the
mirror through the optical cavity.
d2 x Wm Ê dx ˆ Ê 1 ˆ
+ + W2m x = Á (Frp + FL ) (13.13b)
dt2 2Qm ÁË dt ˜¯ Ë meff ˜¯
where x is the position of the movable mirror with respect to
equilibrium, Δ is the detuning of the laser with respect to the cavity
resonance, τ0 is the intrinsic photon decay rate of the cavity, τext
is the extrinsic photon decay rate of the cavity (mainly due to the
mirror loss), Ωm is the mechanical resonance frequency, Qm is the
mechanical quality factor of the mirror, and meff is the effective
mass of the mirror. |a|2 is the optical energy stored in the cavity and
418 Coulomb-Assisted Laser Cooling of Piezoelectric Semiconductors
|s|2 is the optical power incident on the cavity. Fep is the force due
to radiation and FL is the Langevin force, which is responsible for
thermal vibrations of the mirror. According to the above equations,
one can figure out that the movement of the mirror changes the
detuning (i.e., Δ(x) = Δ + (ω0/L)x, where L is the length of the Fabry–
Perot cavity and ω0 is the resonance radial frequency of the cavity),
which in turn changes the optical power inside the optical resonator.
On the other hand the change in the optical energy of the resonator
results in a coherent change in the optical force exerted on the
moving mirror. As we shall see in the following, with a correct phase
difference between the dynamical optical force and the mechanical
motion, an efficient energy transfer between the optical mode and
the mechanical mode can be achieved. To start this analysis, we note
that the optical force exerted on the mirror wall is equal to 2|a|2/
(cTtr) in which Ttr is the round-trip transit time (~2L/c) and c is the
speed of light in the cavity. We expand |a|2 in the weak retardation
regime, that is, 1/t >> Ωm according to Eq. 13.13a to obtain
Ê t 2t / t ex ˆ È Ê 8Dt 2 ˆ w Ê 8Dt 2 ˆ w 0 dx ˘ 2
F (t ) = Á Í1 + Á 2 2 ˜ R x -t Á 2 2
0
˜ ˙s
Ë cTrt 4D 2t 2 + 1 ˜¯ ÍÎ Ë 4t D + 1 ¯ Ë 4t D + 1 ¯ R dt ˙˚
(13.14)
-1
where t = t 0-1 -1
+ t ex
and the first two terms in this series are the
adiabatic response of the force to the changes in position of the
moving mirror. As the mirror moves, the instantaneous detuning
changes which results in the change of optical power and force as
described by these two terms. The second term corresponds to the
mechanical displacement and hence results in an induced optical
stiffness. The power transfer between the optical mode and the
mechanical mode can be found by evaluating F.dx/dt and averaging
it over one period of mechanical oscillation. By doing so, it turns out
that only the third term in the above equation is responsible for a
nonzero cooling/amplification rate because it is in phase with the
velocity. It is also inferred from the above equation that the sign
of the detuning determines the direction of the energy transfer
between the optical mode and the mechanical mode. A red-detuned
laser leads to cooling of the mechanical vibration that would change
the balance in favor of anti-Stokes radiation, as schematically shown
in Fig. 13.4, and the reverse is true for a blue-detuned laser. In
general in any system based on the backaction mechanism, to have
Basics of Optomechanical Cooling and Amplification 419
xw =
ÂF n n ,w
/(1 + iwt n ) + FL ,w w02
(13.16)
K w eff
2
- w 2 + iG eff w
where Fn,ω and FL,ω are the amplitudes of the light-induced force and
the Langevin force at the frequency of ω. τn is the characteristic delay
time of Fn,ω. K is the mechanical rigidity of the mirror (which is equal
to meff W2m). ωeff and Γeff are the effective resonance frequency and
the damping rate of the moving mirror in the presence of the optical
field inside the cavity and can be expressed as follows [26]:
ÊW ˆ È w 0t n Kn ˘
G eff = Á m ˜
Ë Qm ¯
Í1 - Qm
ÍÎ
Âw t
n
2 2
n
˙
+ 1 K ˙˚
(13.17a)
È Kn ˘
Â
1
w eff
2
= w 02 Í1 + ˙ (13.17b)
ÍÎ n
w 2t n2 + 1 K ˙˚
where Kn is the light-induced rigidity and is equal to –∂Fn, 0/∂x, where
Fn,0 is the static light-induced force at equilibrium. Considering the
mechanical dispersion of the moving mirror and spectral properties
of the thermal driving force, one can obtain the following formula for
amplitude of vibration between ω – δω/2 and ω + δω/2:
420 Coulomb-Assisted Laser Cooling of Piezoelectric Semiconductors
2kBT w 02G
<xw >2 = dw (13.18)
p K (w eff
2
- w 2 )2 + (G eff w )2
Intensity Δ1 Δ2
AS
S
S
AS
ω0 ω
ω1 ω2
Figure 13.4 The resonator dispersion and stokes (S) and anti-Stokes
(AS) components of the photons in the cavity. Two different
conditions with negative laser detuning (ω1) and positive laser
detuning (ω2) are shown. In the first scenario the anti-Stokes
component is larger than the Stokes component, leading to the
cooling of the mechanical mode. The reverse is true for positive
detuning, which results in the amplification of the mechanical
motion.
the carriers reach thermal equilibrium, not only do they gain the
thermal energy of kT but also they experience an inherent blue shift
due to the Coulomb interaction. As a result, the difference between
the energy of the spontaneously emitted photons and the absorbed
photons increases, which could be an advantage, according to
Eqs. 13.1–13.3, for increasing the laser cooling efficiency. Similar
behavior has been demonstrated in type II QW structures, where
the excitation of free carriers leads to a blue shift of radiation by the
specific type II band alignment [10]. This type of band alignment
plays the same role as a built-in electric field and separates the
excited electron and hole pairs. This separation directly affects the
Auger and radiative recombination rates. In the next section, the
experimental and theoretical investigations to clarify these effects
are outlined.
e-
ΔEe
ΔEh
h+
8p c
•
(hn )2
(
a (hn ) n0 (hn ) )
(hc )3 EÚ
2
Bni2 = d(hn ) (13.20)
Ê hn ˆ
exp Á -1
Ë k T ˜¯
g
B
where B is the radiative recombination coefficient, ni is the intrinsic
carrier density, α(hν) and n0(hν) are the absorption and the refractive
index, respectively, of the material at a photon frequency of ν, and h
is Planck’s constant. By simplifying the above equation, we obtain
8p c
B= a g ng2(E g + kBT )2 kBT exp( -E g /kBT ) (13.21)
ni2(hc )3
where αg and ng are the absorption and the refractive index at
the average PL wavelength. The above equation shows that at
thermal equilibrium, the absorption and radiative recombination
are proportional to each other. For a QW structure, the interband
absorption coefficient, neglecting the excitonic effects, has been
found to be
a (hn ) = C Â Â
2 2 2
Inm e .pcv d (Enm (k|| ) - hn )( fvm - fcn )
n ,m
V k
||
(13.22)
Effect of the Separated Electron and Hole Wave Functions on the Recombination Rates 425
-5
InAlAs InGaAsP InAlAs
InGaAs InP
-5.5
Energy (eV)
e1
-6
-6.5 hh1
lh1
-7
0 50 100 150 200 250 300 350 400
Position (Å)
Figure 13.6 The band structure and the wave function of the grown
structure used to investigate the Auger recombination process.
The electron wave function is shown by the dotted lines,
whereas the first heavy-hole and light-hole wave functions are
shown by the solid and dashed lines.
Effect of the Separated Electron and Hole Wave Functions on the Recombination Rates 427
Normalized PL intensity
e1-lh
1
0.8 Simulated
0.6 e1-hh
0.4 Measured
0.2
0
1.3 1.35 1.4 1.45
Wavelength (μm)
Figure 13.7 The simulated and measured PL spectra. Two peaks corre-
sponding to the electron–light hole and electron–heavy hole
recombinations were identified and indicated. The simulated
curve is a solid line and the measured curve is shown by the
dotted line.
2.8
T=77 K
2.6
2.4
2.2
T=300 K Auger Slope
2
0.9 1 1.1 1.2 1.3 1.4
Log10 Pump power (mW)
Figure 13.8 The PL peak voltage versus pump power for different
temperatures ranging from 77 K up to 300 K. The Auger slope is
shown by the dashed line and the radiative slope by the dotted
line. The slopes of the measured curves are close to Auger
process, which shows that the carrier density is high enough
for the Auger process to dominate in all of the measurements.
We also note that the best carrier concentration for the purpose
of laser cooling is equal to A / C , where A is the SRH recombination
Effect of the Separated Electron and Hole Wave Functions on the Recombination Rates 429
Measured
-28 Fitted (non-threshold Auger)
10
Fitted (bulk Auger)
-30
10
Structure (cm6/s)
Êd Ê (
Ê 2C 2 ˆ ˆ C Ê uz (0, t ) - uz LQW , t
E g (t ) = CStark Á 31 Á C11 + C12 - Á 13 ˜ ˜ 33 Á
) ˆ˜
Ë Ë Ë C33 ¯ ¯ C13 ÁË LQW ˜¯
(13.25)
where uz is the out-of-plane displacement of atoms within the
width of the QW structure (LQW) and the Cs are elastic stiffness
coefficients. To obtain the above formula, first the in-plane strain
C
component (εxx) was obtained by multiplying the factor - 33
2C13
Ê u (0, t ) - uz ( LQW , t ) ˆ
to the average out of plane strain Á z ˜ [52]. The
Ë LQW ¯
in-plane strain component corresponds to a piezoelectric field
Ê 2d ˆ Ê 2C 2 ˆ
with a magnitude equal to Á 31 ˜ Á C11 + C12 - 13 ˜ xx , where
Ë ¯Ë C33 ¯
d31 is the piezoelectric constant and ε is the electric permittivity
of the medium [52]. The piezoelectric field changes the bandgap
through the QCSE and this change is calculated by including
the Stark coefficient (CStark). It should be noted that for the QW
structure described in here the change of bandgap by this process
is much larger than that of deformation potential (ΔEg,d), that is,
432 Coulomb-Assisted Laser Cooling of Piezoelectric Semiconductors
DE g Ê 2C d ˆÊ 2C 2 ˆ
= Á Stark 31 ˜ Á C11 + C12 - 13 ˜ >~ 10 where D1 and D2 are
DE g,d Ë ( D2 + D4) ¯ Ë C33 ¯
deformation potential parameters.
(a) Phonon
dynamics
Change of Phonon
piezo strain strain
Coulomb
QCSE
interacon Change
Between e—-h+ Of photo-
generaon
Phonon scaering
Electron
Thermal Phonon LO
Bath
reservoir Bath AC Equilibrium at Te
Equilibrium at TL Coulomb
interacon
Non-Radiative
recombination Radiave
recombinaon
Thermal generaon
Figure 13.10 (a) The diagram showing the processes that lead to the
Coulomb backaction mechanism and (b) the energy flow
schematics. The Coulomb cooling increases the acoustic
phonon absorption, as shown by the red flash.
Formalism of Coulomb-Assisted Laser Cooling of Semiconductors 433
and consequently
R
Dn (t ) =
1
Dn0 - d Dn0 /dt (13.30)
Ê wq ˆ
2
R + w q2
2
1+Á ˜
Ë R ¯
d Dn
where Δn0 is the solution of the rate equation when DG . By
dn dt
using Dn0 = DE g,piezo , we can write
dE g
1 Ê dn ˆ R Ê dn ˆ Ê d DE g,piezo ˆ
Dn(t ) = Á ˜ DE g,piezo - Á ˜Á ˜
Ê w q ˆ Ë dE g ¯ R2 + w q2 Ë dE g ¯ Ë dt ¯
2
1+Á ˜
Ë R ¯
(13.31)
Therefore the component of the piezostress, which is out of
phase with atomic displacement due to phonons, will be equal to
1 Ê dn ˆ
s piezo,o = e33 l21 Á ˜ DE g (13.32)
Ê w q ˆ Ë dE g ¯
2
1+Á ˜
Ë R ¯
ÊL ˆ
Icooling,q = 8 Á e-h ˜
Ë arV ¯
1 Ê 1 ˆ Ê 1 ˆ Ê kqz ˆ Ê 1ˆ
CCoulomb ÁË v ˜¯ ÁÁ 2 2 ˜˜ Á k ˜ (1 - cos(kqz Le- h ))ÁË nq + 2 ˜¯ w q
s Ë Le-h kqz ¯ Ë q ¯
2
Ê wq ˆ
1+Á ˜
Ë R ¯
(13.34)
where ρ is the mass density, a is the lattice constant, Le-h is the aver-
age separation of electron and hole wave functions, V is the volume
of the crystal, vs is the group velocity of the phonon mode, kq and kqz
are the amplitudes of the phonon wavevector along the direction of
the propagation and along the crystal growth direction, respectively,
and nq is the phonon occupation number of mode q. CCoulomb is an ef-
fective elasticity and is equal to the ratio of backaction piezoelectric
stress to the average phonon’s strain along the direction of epitaxial
growth and is expressed as
Ê 1ˆ Êd Ê Ê 2C 2 ˆ ˆ C Ê dn ˆ
CCoulomb = Á ˜ e33 l21CStark Á 31 Á C11 + C12 - Á 13 ˜ ˜ 33 Á ˜
Ë 2¯ Ë Ë Ë C33 ¯ ¯ C13 Ë dE g ¯
(13.35)
The obtained formula for cooling per unit area (Eq. 13.34)
shows that the mode with an out-of-plane wavelength of 2.33Le-h
experiences maximum cooling. In Fig. 13.11 the phenomena that
lead to Coulomb-assisted cooling are shown as a function of time for
this particular mode. The change of bandgap and the carrier density
due to out-of-plane strain of the phonon leads to a coherent change
of piezostress, as shown in Fig. 13.11a–d. The component of the
piezostress, which is 180° out of phase with carrier velocity, results
in the net cooling power (Fig. 13.11d–f). The total cooling per unit
area extracted from the phonons by means of Coulomb interaction
is obtained by summing up the contribution of all phonon modes.
The cooling per unit area of phonons modes with a specific energy
2p k||q dk||q k zq
of ћvs(kq)kq is equal to Id3k = Icooling ,q ¥ 3
, where L is
Ê 2p ˆ
ÁË L ˜¯
the length of the crystal and k||q is the magnitude of the in-plane
wavevector. Thus summing up Id3k over the whole Brillion zone
436 Coulomb-Assisted Laser Cooling of Piezoelectric Semiconductors
Ê Ê 2p ˆ Ê p pˆ
Á k||q ŒÁ 0, ˜ and k zq ŒÁ - , ˜ [34] gives the net cooling per
Ë Ë a ¯ Ë a a¯
unit area:
p Ê 2p ˆ
kzq = k||q =Á ˜
a Ë a ¯
Ê 2 ˆ
Icooling =
p
ÚÚ Á 2 ˜
Ë ap r ¯
kzq =- ,k||q =0
a
Ê 1 ˆ1 Ê 1ˆ
CCoulomb Á ˜ (1 - cos(kqz Le-h ))ÁË nq + ˜¯ k||qdk||qdk zq
Ê w q ˆ Ë Le-h kqz ¯
2
2
1+Á ˜
Ë R¯
(13.36)
Δεzz,max 0
ΔEg,max 0
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
Δno,max 0
Δσo,max 0
0 2 4 6 8 10 12 14 16 18 20
0 2 4 6 8 10 12 14 16 18 20
(e) (f)
Δvavg 0
Pcooling 0
Figure 13.11 The time diagram of (a) the average out-of-plane strain, (b) the
change of bandgap due to the change of piezoelectric field, and
(c) the change of carrier density due to the change of bandgap.
It is composed of an in-phase (Δni) and an out-of-phase (Δno)
component. The amplitude of the out-of-phase component is
5.4 × 10–5 times lower than the total carrier density change
for the phonon mode under consideration. (d) The piezostress
due to Δno and (e) the average velocity. It is in phase with the
piezostress shown in (d) and therefore results in a net cooling
power, as shown in (f).
We can then simplify the cooling per unit area by defining the
function F:
Formalism of Coulomb-Assisted Laser Cooling of Semiconductors 437
p Ê 2p ˆ
kqz = ,k||q =Á ˜
a Ë a ¯
Ê 1 ˆ
ÚÚ
1
= Á ˜
v2s (kqz
2
+ k||2q ) Ë Le -hkqz ¯
kqz =0 ,k||q =0 1+
R2
Ê ˆ
Á ˜
Á ˜
Á ˜
1 1˜
(1 - cos(kqz Le-h ))Á + k||qdk||qdk zq
Á
Á
Ê v
Á s
(k 2
qz + k||2q ) ˆ
˜
2˜
˜
Á exp Á
kT ˜ -1 ˜
Á ÁË ˜¯ ˜
Ë ¯
(13.37)
and then express Icooling as
Ê 2 ˆ
Icooling = Á 2 ˜ CCoulomb (13.38)
Ë ap r ¯
Therefore the cooling efficiency can be calculated as
Ê ˆ
Á ˜
Icooling Á 2 ˜
hcooling = =Á Ccoulomb
Iabs ˆ˜
Á ap 2 rI(1 - exp( -a L ) + a 2PA LQW I ˜
ÁË QW
1 + a 2PA LQW I ˜¯ ˜¯
(13.39)
A wurtzite c-GaN/InGaN QW, as shown in Fig. 13.12, was
considered to explore the Coulomb effects. The k.p method in
conjunction with Poisson equation can be used to calculate the
self-consistent energy levels and the absorption of the MQW [52]
in presence of Coulomb interaction. Effective masses and band line-
ups were calculated according to Ref. [9]. The QW structure consists
of 1 nm of In0.15Ga0.85N sandwiched between 6 nm of GaN barriers.
As can be seen in Fig. 13.12, the spatial separation of electron and
hole wave functions due to the presence of a large piezoelectric
field, leads to the screening of the built-in and piezoelectric electric
field after photogeneration. As a result, significant blue shift of the
interband transition energies is expected for a large photogenerated
carrier density due to the QCSE.
438 Coulomb-Assisted Laser Cooling of Piezoelectric Semiconductors
-3
GaN In0.15Ga0.85N GaN
-3.5
-4
Energy (eV)
3.212
-4.5
Ee-Ehh (eV)
3.21 Phonon’s stress
-5
3.208
-5.5
3.206
1 2 3 4
-6 x 10
19
Piezo stress
n(cm-3)
-6.5
-7
-7.5
0 20 40 60 80 100 120 140
Position (Å)
Figure 13.12 The band diagram of the designed MQW system. The electron
wave function is in green, whereas the heavy-hole wave
function is in red color. The inset shows the bandgap versus
carrier density. Note that the direction of piezostress will be
reversed for a phonon with opposite strain. This results in an
effective phonon damping.
The inset of Fig. 13.12 shows the bandgap versus the carrier
density with the slope equal to Le–hλ21. According to Fig. 13.13a, we
can see that the net cooling efficiency increases significantly with the
presence of the Coulomb interaction. We can see that the net cooling
is possible for a laser intensity of 5 mW/μm2 and for the detunings
between 0–4 THz. The dependence of the net cooling efficiency on
laser intensity can be explained by the fact that both the external
quantum efficiency ηext and the Coulomb cooling are dependent on
laser intensity. At very high intensities, the probability of two-photon
absorption increases, which is almost independent of detuning near
the band edge [23] . As a result, in this regime, the change of carrier
density with the change of bandgap reduces and the Coulomb
cooling efficiency is suppressed. This behavior can be understood
considering that the Coulomb cooling efficiency decreases due to
the reduction of dn/dEg. For extremely low excitation levels, the SRH
recombination of the photogenerated carriers reduces the external
quantum efficiency and the net cooling. The reduction of the internal
quantum efficiency for large blue detunings is due to the increased
carrier density and domination of the Auger recombination rate. In
both cases, the reduction of the internal quantum efficiency effectively
diminishes the positive effects of the Coulomb interaction on the net
Formalism of Coulomb-Assisted Laser Cooling of Semiconductors 439
laser cooling. As a result, below and above certain detunings, the net
cooling efficiencies for both cases (with and without the Coulomb
effect) become essentially equal.
(a) cooling
0
With Coulomb
-20 Without Coulomb
ηc,net (%)
5î 10 3 mW/μm2
-40
5 mW/μm2
-60
-100
-20 -15 -10 -5 0 5 10 15 20
Detuning (THz)
(b)
-4
10
only anti-Stokes
anti-Stokes with
Coulomb cooling
-6
τnrb (s)
10
30 meV
-8
τnr=7.8×10-8 s 19 meV
10
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
ηe
Figure 13.13 (a) The net cooling efficiency versus laser detuning for various
laser intensities. Cases with both anti-Stokes and Coulomb
cooling are shown by the dotted lines. The solid lines show only
anti-Stokes cooling. These simulations have been performed
for the structure that has an absorption broadening of 19 meV
when the entire PL is extracted out from the material without
reabsorption. (b) The break-even nonradiative recombination
lifetime (τnrb) versus extraction efficiency for only anti-Stokes
cooling (solid line) and both the anti-Stokes and Coulomb
cooling (different values of absorption broadening has been
considered). The 30 meV absorption broadening is a usual
experimental value [19]. The lowest broadening is expected to
be close to 19 meV (the value for GaN epilayers [15]). The red
line represents the nonradiative recombination lifetime for a
similar structure [53]. In all of the simulations, it is assumed
that the temperature is 300 K. Reprinted with permission from
Ref. [30]. Copyright 2014, AIP Publishing LLC.
440 Coulomb-Assisted Laser Cooling of Piezoelectric Semiconductors
Broadened excited
Level due to collisions
Quartz E2
window
fluorescence
E1
Carrier fluorescent
atoms atoms
Figure 13.14 (a) The schematics of a gaseous system that consists of active
florescent atoms and carrier atoms. The collision between
the atoms results in broadening of the luminescence and
upconversion of the luminescence energy with the respect to
the absorbed photon energy. (b) The energy level diagram and
the excitation and radiation of an atom in the system. With
collisions the atoms can be excited to the upper state and emit
photons with a center frequency equal to (E2 – E1)/ħ. Therefore
a red-detuned excitation results in cooling.
Energy
Excited
Fluorescence
excitaon
Ground
d0 Inter-atomic
distance
Figure 13.15 The energies of the ground state and the excited states versus
the interatomic distance. The average interatomic distance
is denoted by d0. This graph is similar to the one indicated
in Ref. [49]. This technique of cooling is similar to Coulomb-
assisted cooling of piezoelectric materials in the sense that
in both cases the photoexcitation happens as the interatomic
distance decreases and the fluorescence is a slower process
that extracts the excess energy of the carriers outside of the
material. However, the mechanism of a shrinking bandgap and
the excess energy gained by the carriers is different for both
cases.
Concluding Remarks 445
Acknowledgments
The authors would like to acknowledge partial support from NSF
awards nos. ECCS-1310620, ECCS-1206155, and ECCS-0901855, as
well as ARO award nos. W911NF-13-1-0485 and W911NF-11-1-0390.
The authors would also like to acknowledge the help of Travis
Hamilton, Eric Dexheimer, and Abigail Turay in preparation of this
chapter.
References